diff --git a/.gitattributes b/.gitattributes index a6344aac8c09253b3b630fb776ae94478aa0275b..52373fe24473b1aa44333d318f578ae6bf04b49b 100644 --- a/.gitattributes +++ b/.gitattributes @@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text *.zip filter=lfs diff=lfs merge=lfs -text *.zst filter=lfs diff=lfs merge=lfs -text *tfevents* filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text diff --git a/config.json b/config.json new file mode 100644 index 0000000000000000000000000000000000000000..e88ef5e29e607aa30cad6553db97b8d855af8ea6 --- /dev/null +++ b/config.json @@ -0,0 +1,38 @@ +{ + "_name_or_path": "Nemotron-4-340B-Base-hf", + "architectures": [ + "NemotronForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 2, + "eos_token_id": 3, + "head_dim": 192, + "hidden_act": "relu2", + "hidden_size": 18432, + "initializer_range": 0.0063, + "intermediate_size": 73728, + "max_position_embeddings": 4096, + "mlp_bias": false, + "model_type": "nemotron", + "norm_eps": 1e-05, + "num_attention_heads": 96, + "num_hidden_layers": 96, + "num_key_value_heads": 8, + "quantization_config": { + "activation_scheme": "dynamic", + "ignored_layers": [ + "lm_head" + ], + "quant_method": "fp8" + }, + "rope_percent": 0.5, + "rope_percentage": 0.5, + "rope_scaling": null, + "rope_theta": 10000, + "tie_word_embeddings": false, + "torch_dtype": "bfloat16", + "transformers_version": "4.43.0.dev0", + "use_cache": true, + "vocab_size": 256000 +} \ No newline at end of file diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000000000000000000000000000000000000..48aaabbb180a6c12eec92ff6a9dc11edb619bbe1 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,6 @@ +{ + "_from_model_config": true, + "bos_token_id": 2, + "eos_token_id": 3, + "transformers_version": "4.43.0.dev0" +} diff --git a/model-00001-of-00189.safetensors b/model-00001-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..99515d3c6bf3e960135cfa2af739745b44accfff --- /dev/null +++ b/model-00001-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:492e67d00b85927fd5106a5fcc61933a93a4da2ba3dab19ad84df45908d22f2c +size 9437184112 diff --git a/model-00002-of-00189.safetensors b/model-00002-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2f7f0a5aaa37587d812a9d7f8da61874013cb82b --- /dev/null +++ b/model-00002-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:206eb1589c0ec52919a0bd5fb6bba6454cc0a5b62c987675b1b5c2e0a9551cf0 +size 2095055948 diff --git a/model-00003-of-00189.safetensors b/model-00003-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c8a62589870d4e88bddc59ee21ceb3513d4403c0 --- /dev/null +++ b/model-00003-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e22dc1406f95a7cc26fb7ccb637142678fe94c45d818ee9d7c23e95b885d4a54 +size 2095203836 diff --git a/model-00004-of-00189.safetensors b/model-00004-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..22c52d15adb5f144f6ddf7c89520bae852e1e32c --- /dev/null +++ b/model-00004-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cb7e046722c5d7b8453d84df84c9d2b37a6963c55d470d77ed6c193a470e494 +size 1358954708 diff --git a/model-00005-of-00189.safetensors b/model-00005-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1b45700867a0c05df9aea1a7ac0af06ca8ca40b8 --- /dev/null +++ b/model-00005-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a6fc66b6dc8d3bc39fbd88ffd670ede2b54d32151f7a344908de89223b14a12 +size 2095203836 diff --git a/model-00006-of-00189.safetensors b/model-00006-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ec1becbb6c2f852de84d8237a6ac4284455e8aab --- /dev/null +++ b/model-00006-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa95ca271a1119e2e6ff98c86a50672a110a29ccb14bb5ed01c637813a63b20a +size 1358954708 diff --git a/model-00007-of-00189.safetensors b/model-00007-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..586da9c497698ba5184946d0dcd937d32ec394f9 --- /dev/null +++ b/model-00007-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a3a0f5447817fda87bac0566c5a0061cd445acb79529d6f781da05c8863f2297 +size 2095203836 diff --git a/model-00008-of-00189.safetensors b/model-00008-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4e735fbc88f8a62902369761dc6dc2c7a5e2934e --- /dev/null +++ b/model-00008-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f5da863a3e6ad19422f9f75d9e6f9a4b279053185201298974b9ec6ac0dcbb5 +size 1358954708 diff --git a/model-00009-of-00189.safetensors b/model-00009-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ad780b60e8705e6d58f39fc7beb1e1ebb814ecca --- /dev/null +++ b/model-00009-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8d53665f8e452a4cc6b670304c9f9acb2ab3fc96912768b8267d9182c08c01f +size 2095203836 diff --git a/model-00010-of-00189.safetensors b/model-00010-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b40846a44e2fec0ac639adcf8140ba24b44d8a94 --- /dev/null +++ b/model-00010-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4c94783835bfe7e3b999687b98925c05cd0d38febc74ce8129820813731fddf +size 1358954708 diff --git a/model-00011-of-00189.safetensors b/model-00011-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b3ffdc03160606852b16f77f2c291bec1d87ffa5 --- /dev/null +++ b/model-00011-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:941968406fabb8602f1145d9e0e728780f83be55c85bdb9420319b943d83a749 +size 2095203836 diff --git a/model-00012-of-00189.safetensors b/model-00012-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..72dc466097e6c2aaf0061754b881fa9f9703fcd3 --- /dev/null +++ b/model-00012-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b45a29bc8265bd68003ec4c6cd9d967456a61d36b22087fc290368ccb7b4ab1c +size 1358954708 diff --git a/model-00013-of-00189.safetensors b/model-00013-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..daa444afd1190499808ac001c887706265cf669e --- /dev/null +++ b/model-00013-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5f3c588f6c289165d22def9abdde4af230b78cd5697ce962593d3ffb38486c4 +size 2095203836 diff --git a/model-00014-of-00189.safetensors b/model-00014-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..539b06251fe9c9572a543dc2a0949e716c133090 --- /dev/null +++ b/model-00014-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:197befcef18e6f1ee2634db13ae3f114aff3fd6d32979cd15f7db3ce1361cd9e +size 1358954708 diff --git a/model-00015-of-00189.safetensors b/model-00015-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e2259892d571fadec8b8b25b1eb530849fa2cea9 --- /dev/null +++ b/model-00015-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96540f02c449ad300d9764a9bc98698fef35d8b374e6b35d2709ed7fa04a0364 +size 2095203836 diff --git a/model-00016-of-00189.safetensors b/model-00016-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1c098524a9dfe2ab93406b086c1f602a562e3379 --- /dev/null +++ b/model-00016-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6b33f3c82e2462a3852fc80459c27b5eda495cbf7efdb5fd384dd6ffec8aa7c +size 1358954708 diff --git a/model-00017-of-00189.safetensors b/model-00017-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..398900f1697f9cbfab7296774673cea02fe8db76 --- /dev/null +++ b/model-00017-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dcee331e8c683d4d5bcec6568d96360221496d3588cd5d255e4401e847938bb5 +size 2095203836 diff --git a/model-00018-of-00189.safetensors b/model-00018-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3c7cf3ab0e26f7423508a66014f6a830839a4ee7 --- /dev/null +++ b/model-00018-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2dedde9514dcd9b03a33f1b6f4bf54c14e16ee6d14a6a01bcfdb7b4864b239b5 +size 1358954708 diff --git a/model-00019-of-00189.safetensors b/model-00019-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0ec1445b08d784683ad67061b9135c238238f42a --- /dev/null +++ b/model-00019-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:132daf633959e27797a12924f82c5079f76018a1ecf1f26a10a0f0eabf8d3442 +size 2095203836 diff --git a/model-00020-of-00189.safetensors b/model-00020-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5e4745c15df959345cea2ce587ebc2397390441c --- /dev/null +++ b/model-00020-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78fbe48f9aadff8af34d7d9761ddb5d84ea00d640b788dfa9f82af36cbd45ba0 +size 1358954708 diff --git a/model-00021-of-00189.safetensors b/model-00021-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..76b3bb216c4cd7166ced7ad02c613513f0aa0aed --- /dev/null +++ b/model-00021-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5af142a24bedec15cbfaf0ace484cbdf9ff01a2b4b99f72a85c98e5dddea145 +size 2095203836 diff --git a/model-00022-of-00189.safetensors b/model-00022-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..94b3ea560d89052d498083fa5cf5bd7d76bdf2a7 --- /dev/null +++ b/model-00022-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:679060a348a979ff966fbf89fccceec278b0176b58199395b659ef6001324a0d +size 1358954716 diff --git a/model-00023-of-00189.safetensors b/model-00023-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9353d5c576817f08af9b472ed06b2cc619501162 --- /dev/null +++ b/model-00023-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c02f136ad71ba9947022ffee91b71a7bdf60d1a72111a0fe681e6a6ad9364ec7 +size 2095203852 diff --git a/model-00024-of-00189.safetensors b/model-00024-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6d7943f943feef9d38de80e9e6eab4967e661bd0 --- /dev/null +++ b/model-00024-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac2e54c71d5a7fb7e214c1121c7737fd495ee1cdd7cbbfe22a1013b3ad3fbfdd +size 1358954716 diff --git a/model-00025-of-00189.safetensors b/model-00025-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..43dd2726932d58f054c17c3e6795250bfa3170bc --- /dev/null +++ b/model-00025-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4cc4137de93bf43bd726f822d5ecc0fb3462cf887ba0de4d327c075aa31703c0 +size 2095203852 diff --git a/model-00026-of-00189.safetensors b/model-00026-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d40881400f45ed66386783de4dbf8455081941eb --- /dev/null +++ b/model-00026-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2990fdbcd669429ae8cb740013755ffb02b6be9600e2eff431621bd005a9164 +size 1358954716 diff --git a/model-00027-of-00189.safetensors b/model-00027-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8e23d9aaf01a8e315e3903ab6f191740da7f2030 --- /dev/null +++ b/model-00027-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94c7e42d62da64c135874372113cdcfc67398867222883b8a22c5ae72dc83606 +size 2095203852 diff --git a/model-00028-of-00189.safetensors b/model-00028-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..20662b67ee4ee65b2f0c98aa13ee886d872214ff --- /dev/null +++ b/model-00028-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b5eee630e2f6f7a16a610d6811cb640c1ff61442a1ccbfc0fd48644ccdf3c5e +size 1358954716 diff --git a/model-00029-of-00189.safetensors b/model-00029-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..75991516f3f6c35cf7439c3f40457c4ca4cabd6f --- /dev/null +++ b/model-00029-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d93081ad2411540e79148032d85f521c5fa1c22b63b19d36ba373811fd549ad1 +size 2095203852 diff --git a/model-00030-of-00189.safetensors b/model-00030-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4a84a15adaa3ce41120a5303a2fedb7302770105 --- /dev/null +++ b/model-00030-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:749ac01f582ca0c981e8b3afa20c5fae32b3022e426054d179ff002a97928a6c +size 1358954716 diff --git a/model-00031-of-00189.safetensors b/model-00031-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..89c35fe557cb27b9243a3452f42be5300b5474cf --- /dev/null +++ b/model-00031-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f96672bb43116626c47df7d139dbc5fb9cfe7a56bfaf5d90c9ae7a60c845b1b +size 2095203852 diff --git a/model-00032-of-00189.safetensors b/model-00032-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..855a031a30353e74aae1cb7bca3b516f6c908b59 --- /dev/null +++ b/model-00032-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c5cfaa532708b5befa6145eeb37a0ad761cb8d103fd316ccc5b7404731b77ef +size 1358954716 diff --git a/model-00033-of-00189.safetensors b/model-00033-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2301c1756cf234344c33bae15740d62342b7241d --- /dev/null +++ b/model-00033-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b742f787298466e033701ab75be0616836ba85fad05a341482cd801297416593 +size 2095203852 diff --git a/model-00034-of-00189.safetensors b/model-00034-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..600bb9b0dde4e80f6ba51a655fdd34e6bd6965be --- /dev/null +++ b/model-00034-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75f57a0c96f25c1f6c58a1a7739226bdf12ef6a17a713900a7e6dfa81aa748fd +size 1358954716 diff --git a/model-00035-of-00189.safetensors b/model-00035-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c5c4d732336cb23fa7fa0797cfad72ca07f434ea --- /dev/null +++ b/model-00035-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cbd533f9e874956f922a85535b37303009dd6254f6c014f376f4c7470ab17c4 +size 2095203852 diff --git a/model-00036-of-00189.safetensors b/model-00036-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9ef9405778edf4c43bb7fe625c886dfd1ca68a4f --- /dev/null +++ b/model-00036-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4be0f31cac813940a69778066be18a8565b094734c73a5fc54749634ab3e1468 +size 1358954716 diff --git a/model-00037-of-00189.safetensors b/model-00037-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..45747ac0e4bbb53ce173b06a3905dbb6fcc54084 --- /dev/null +++ b/model-00037-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:57e5a9c28a897b2ac80321a1053c02cf610f0d1895ba72ffba47fa4643f9e556 +size 2095203852 diff --git a/model-00038-of-00189.safetensors b/model-00038-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b5ecb1a71a3db365a6a08dec31028f55a11ff287 --- /dev/null +++ b/model-00038-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce7b84d7e0ede36de80d854d4140cfb8c0f66b5af4977322ec8ade864ae6b964 +size 1358954716 diff --git a/model-00039-of-00189.safetensors b/model-00039-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4f138a158c230fe4d761e8bfcf4f433f7ec3520f --- /dev/null +++ b/model-00039-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f85733ece4a869c5e4b50647000e04c83ab5867cc0f152165504c1d15b321555 +size 2095203852 diff --git a/model-00040-of-00189.safetensors b/model-00040-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d83dbfda3623ef24bd67e4569da52bac04e265ce --- /dev/null +++ b/model-00040-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d71be466b49c450f5bb49cef853bc0d70e9713c4b88178297d3847bae631808 +size 1358954716 diff --git a/model-00041-of-00189.safetensors b/model-00041-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f5e0dd8e81a8f0b1bf7f35bfd1e936d919014889 --- /dev/null +++ b/model-00041-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c2efd9d258014378e422ead20ce17e50c3204822092bc9bbc0673fb521057d9 +size 2095203852 diff --git a/model-00042-of-00189.safetensors b/model-00042-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d6d4ef6bd229b13f27924fae9921a532088aeb86 --- /dev/null +++ b/model-00042-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0176156bd2ec62eab16e4568e1b2bdb4ecd8e29c6ae09975b254fd439f86e4cc +size 1358954716 diff --git a/model-00043-of-00189.safetensors b/model-00043-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fe74c7540f6fcea092794cb63f7b4ddebd9072bd --- /dev/null +++ b/model-00043-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:47e072411cf28b82cbc9805fb81a8ee9e4e6480c585f2d4bd25d0268dfd831f1 +size 2095203852 diff --git a/model-00044-of-00189.safetensors b/model-00044-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4ebbc5968695e4aae3d13bd3a4c410a3a2dea401 --- /dev/null +++ b/model-00044-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aef2efdcda387a1c9e1b2e16b5d34258fbe1970fc17ede4adec7e2d43e640676 +size 1358954716 diff --git a/model-00045-of-00189.safetensors b/model-00045-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b132926dc18a46cc6d1351c21a34b1acd1517ba7 --- /dev/null +++ b/model-00045-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64f42af2ad24db779bc5866943455c9c53310db8817ec2244edcb04cb84f562f +size 2095203852 diff --git a/model-00046-of-00189.safetensors b/model-00046-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d9ce82743a4f3500351cc43137971d359fe06b34 --- /dev/null +++ b/model-00046-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea9be991f8da98514148b6eca5fe6a66c35775ca2204dc8723c0e9b29c7be89f +size 1358954716 diff --git a/model-00047-of-00189.safetensors b/model-00047-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..79ed6a8ab0a7ace3a82a28b75e3388938e4e6acc --- /dev/null +++ b/model-00047-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86c9fa06f9b0bd1d3b092361d662208be9c11b157a13f7c600ddb84f97b925dc +size 2095203852 diff --git a/model-00048-of-00189.safetensors b/model-00048-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f24a8a351ab4fc31eab6435cc19b6732001fcc2c --- /dev/null +++ b/model-00048-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac65f422979ed9fa1a8acc614d2ec24350f7bf31cc732667e43bed6024244c94 +size 1358954716 diff --git a/model-00049-of-00189.safetensors b/model-00049-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c058acd919933f48bfe6585754bda5eed135264b --- /dev/null +++ b/model-00049-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b582e5e29f6695db91e646cab8aa19ef8887b75e748f78f92e65ffa67fcaf763 +size 2095203852 diff --git a/model-00050-of-00189.safetensors b/model-00050-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2980027cf8917d282ee85eadcf368bd0c2c04da9 --- /dev/null +++ b/model-00050-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ce51c01c4d6fa3a1eb751188bd59244264a45899ac275a13a54c229e3f646e54 +size 1358954716 diff --git a/model-00051-of-00189.safetensors b/model-00051-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..91126de67b3b81bb4e6e5c61dc9e562907765985 --- /dev/null +++ b/model-00051-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9046bd2ea890ef97c13b51adb6b5265570e32772c418901c6fcef14a96e78c19 +size 2095203852 diff --git a/model-00052-of-00189.safetensors b/model-00052-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c039b7ce4ad1881946998cb157eafd331dd37eea --- /dev/null +++ b/model-00052-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eb13f8ca89ab69400f8f0618da9b2141c0027dc5c4882c949f91ccf5203da206 +size 1358954716 diff --git a/model-00053-of-00189.safetensors b/model-00053-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8288e15ab92c1925a5c2790f281dd138a79ab549 --- /dev/null +++ b/model-00053-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6646e31b1a7e6eb76a9faa8549d0ce642468fe24a0399b1e82653985d3898666 +size 2095203852 diff --git a/model-00054-of-00189.safetensors b/model-00054-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0d7a62a11724f13ab7cafcdbdd4ba893197ffa43 --- /dev/null +++ b/model-00054-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e4cbe63aa4b1873b09e750bac2f163b1fa042f1161a3f15b4c84a6a6bbd0a2b +size 1358954716 diff --git a/model-00055-of-00189.safetensors b/model-00055-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..dd1e828867339905bebbc798aaf16254d4d326b5 --- /dev/null +++ b/model-00055-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2905b3b6c062f7e4fe32e2a86c2920a289ac0a130cc950b512e38668f793eb37 +size 2095203852 diff --git a/model-00056-of-00189.safetensors b/model-00056-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c461d1656b052e8b59235f8d3efffff249b5da3a --- /dev/null +++ b/model-00056-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab010a95cdaff79169c32ae203b49a221ac6beba6d246f6085d607390323ba11 +size 1358954716 diff --git a/model-00057-of-00189.safetensors b/model-00057-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8e67cbede209f2e9afcbfee427619f94afb444a1 --- /dev/null +++ b/model-00057-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f87d49514f7ea5a16cf1fa54d868fe2df885d5b557a8b65d0fe0e1eaad9b1f8 +size 2095203852 diff --git a/model-00058-of-00189.safetensors b/model-00058-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..989f5315e27df602038209e0a78074a3c4708997 --- /dev/null +++ b/model-00058-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff7a71e2466e7361cd4fabd03da5ba026e49b4144c7630c8465ad59f1a954497 +size 1358954716 diff --git a/model-00059-of-00189.safetensors b/model-00059-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f23298a95ad8dc24d77bff16ee04b8c6765acf40 --- /dev/null +++ b/model-00059-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e6612db46e8d2a625f22390720157aebb15008cffcb29de5c715b4687ffc826 +size 2095203852 diff --git a/model-00060-of-00189.safetensors b/model-00060-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6b51658ce51c6ec3add5201eb8a412b0650f403c --- /dev/null +++ b/model-00060-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:536641b780c1eabcdcd5c58711cac0739d30cfdf2d7893535af3bf7bf9c04bd7 +size 1358954716 diff --git a/model-00061-of-00189.safetensors b/model-00061-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c0897219d86c6e4c020723235ee3f514d9333e68 --- /dev/null +++ b/model-00061-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:09c755e06c3b68441676a324a3f5ce2c29bcfc7b249aa2323f9f7efa1cd1e46e +size 2095203852 diff --git a/model-00062-of-00189.safetensors b/model-00062-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..16befec292c516f5309e1ae098a63d020b0251cb --- /dev/null +++ b/model-00062-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:892d1d04bd9e01e779973ccb0e97c75c703c896b6830d4aeef8cbca6c7e1eafd +size 1358954716 diff --git a/model-00063-of-00189.safetensors b/model-00063-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0f1124bc916cc9613cfb9e1882e1263802f14cac --- /dev/null +++ b/model-00063-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a8ba9ebc456f4786e6f492f0c0a7d8b8f6c92315657e55d5c613b0a06b0e0a6 +size 2095203852 diff --git a/model-00064-of-00189.safetensors b/model-00064-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0f3bac63e73a66f6aeee5a8b6af290b7ceb5dd24 --- /dev/null +++ b/model-00064-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:376ebc2036987cea359adadb066d87f6dda5add17cdc987ec02b988c08fd480a +size 1358954716 diff --git a/model-00065-of-00189.safetensors b/model-00065-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1a08aa86aac4823cff1f5dd4d9ed2cdb9c8a19e8 --- /dev/null +++ b/model-00065-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c4e72b9b30141bdaf019304a96a5fb131f2ddf9933ef0b37c354379ca700ecb +size 2095203852 diff --git a/model-00066-of-00189.safetensors b/model-00066-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..800e62c189f104e02ee78bd61752507b883c3da1 --- /dev/null +++ b/model-00066-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c4a9c77b8df59e6838b88c3a35bf0a4db7e3b46d83f3ff214bd3e48a923ec19 +size 1358954716 diff --git a/model-00067-of-00189.safetensors b/model-00067-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..05fda2083c4394dff0030173b4de783dd8091464 --- /dev/null +++ b/model-00067-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb9408e834c11e3fceb156ec9053982f7d42329055f983eed0520d7b8d68c084 +size 2095203852 diff --git a/model-00068-of-00189.safetensors b/model-00068-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d9d98c1e790a76e7e6bb53ada70abedb85b5f5b6 --- /dev/null +++ b/model-00068-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efcb5a1278fb844f56829901dc7595dc85d88e3b22f46547d468965b33022242 +size 1358954716 diff --git a/model-00069-of-00189.safetensors b/model-00069-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2f35672e8acb9d45e0575d3c76abbde941d6b750 --- /dev/null +++ b/model-00069-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:312844e3f5cfdee83bdd729983816230822f53b166607e74cdace2ca6347b92a +size 2095203852 diff --git a/model-00070-of-00189.safetensors b/model-00070-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5dcabe91ef857c5a859a3337e0634c1027a3405e --- /dev/null +++ b/model-00070-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ab82148e6e964d4f0e9f4ba7a3d9febffffc1d9fd897c4e580c1ca55ae416de +size 1358954716 diff --git a/model-00071-of-00189.safetensors b/model-00071-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5b610ce79713216e7dd90eba362cfa649ae08daf --- /dev/null +++ b/model-00071-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf39a2785ff9b59bdb17a5ecec3c98771aab1e4ce973b81fcd77dfce77983ec0 +size 2095203852 diff --git a/model-00072-of-00189.safetensors b/model-00072-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8544292f1976fa3a8186bbb4a024e35805239e44 --- /dev/null +++ b/model-00072-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:518fe2659d542848aed6f7077acff6c102ea435f7612af7d682648192250d8d6 +size 1358954716 diff --git a/model-00073-of-00189.safetensors b/model-00073-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..685671ae4b92271e75102ce9a0c0dfa26d73a58d --- /dev/null +++ b/model-00073-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e856f9474a1431d4397f3cd7be7750f6919799d90d6d979d180a9d745341f41 +size 2095203852 diff --git a/model-00074-of-00189.safetensors b/model-00074-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e6caa8f3a36d778b1d6a30f22f292cde685cf154 --- /dev/null +++ b/model-00074-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:feab952b8ca91ac96f46bb9502fb4343c8efeceed7473f86a6ee23ce358a26bc +size 1358954716 diff --git a/model-00075-of-00189.safetensors b/model-00075-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9cb8baa622d8d607cc064569053d4817d55c9d93 --- /dev/null +++ b/model-00075-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a07cc72d688cca2970951af778a5699c9bd5ada399303bee3580bbfd4ca1d31 +size 2095203852 diff --git a/model-00076-of-00189.safetensors b/model-00076-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..de10332314cd98efeab02fbcd977eecdabb3ba04 --- /dev/null +++ b/model-00076-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af915f28a4685338d3ae5060437070206e407e79af9e3cd02d0166f84ce8229a +size 1358954716 diff --git a/model-00077-of-00189.safetensors b/model-00077-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7613de2bd622d01358a48000f4a2fab6cc1a835c --- /dev/null +++ b/model-00077-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54f545fe7a40754ead0acec7af34fe4036de413f11875fc0f877fd3c14d207c8 +size 2095203852 diff --git a/model-00078-of-00189.safetensors b/model-00078-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f51391855ca8e25d3e8fae02b42a5e1ddff2e98d --- /dev/null +++ b/model-00078-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32a35f0e7149da900909a132569ea133d7e284548e417f8981b26b103dd1162f +size 1358954716 diff --git a/model-00079-of-00189.safetensors b/model-00079-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..66bf5f6e0fb1aededbf8677693189fb45741ea6c --- /dev/null +++ b/model-00079-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b0261b0410697f6e8f9a2e738c586869c6ecd06d21de16b096cde1d0f4eba43 +size 2095203852 diff --git a/model-00080-of-00189.safetensors b/model-00080-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e62b7caa2ebd8fca285c943befc1a53ed327e656 --- /dev/null +++ b/model-00080-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4c2333d7b60616abe3d6823fd8ed3f57835fe64ab7d0528a3dbe18277004564 +size 1358954716 diff --git a/model-00081-of-00189.safetensors b/model-00081-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..dd5c815434658ff4a6247ca8143ed3f096d6d6ff --- /dev/null +++ b/model-00081-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2abfc91f592dce1e28c49051d4a0dd54fa90414e94c6b09bd7293f3dd6c7e82 +size 2095203852 diff --git a/model-00082-of-00189.safetensors b/model-00082-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..36a43e2da07e1fb293a108460e5a44a24c0f51dd --- /dev/null +++ b/model-00082-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f625c6e82f43aa4183434df3b8f2e0d0dd45be8f5db94ffe5e019c778fd7b4c4 +size 1358954716 diff --git a/model-00083-of-00189.safetensors b/model-00083-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c56af272acb7f088b8e42e136fbb9163b48f01bd --- /dev/null +++ b/model-00083-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25076c0b4c0651ec40595c1f9d4bef9fbbe9e6290823d9c22ba1390149304434 +size 2095203852 diff --git a/model-00084-of-00189.safetensors b/model-00084-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bc43b46e0b57af7618b9de115be586f17a2cade3 --- /dev/null +++ b/model-00084-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0988d4fa3c23425cc786c61b256dbb4414125052e600f4d30dfa3af0b481539b +size 1358954716 diff --git a/model-00085-of-00189.safetensors b/model-00085-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..426aaf1b7c4edd1cc5d878fc1db5838232619c5d --- /dev/null +++ b/model-00085-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f5f806da8a8337fcc6670b01d5dd36c64d82ebb10f8f0e8a4211ff7f56a9a86 +size 2095203852 diff --git a/model-00086-of-00189.safetensors b/model-00086-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6b827bc7caa7265613951ec7383373892af26624 --- /dev/null +++ b/model-00086-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9fbac6c1f149cba08dc1887a010d527f330d47e3904d81f1297239d6e1c7ae91 +size 1358954716 diff --git a/model-00087-of-00189.safetensors b/model-00087-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bf8bf0c43c97fe131b69a4c61c671a3d5e71e20d --- /dev/null +++ b/model-00087-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf8bb67a74e6af435670aa491210cd117ddc28d4ba69fb671b5ee1e791d41ad3 +size 2095203852 diff --git a/model-00088-of-00189.safetensors b/model-00088-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0a4f5451a253694b32e1c72c7e669737645651b7 --- /dev/null +++ b/model-00088-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0556d88cd682858aa48559320ae865cc679aeffe69c41b224a2de7d6363e6bb7 +size 1358954716 diff --git a/model-00089-of-00189.safetensors b/model-00089-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..87c89aab870a9a28730d2de0ddddfccd38e785b3 --- /dev/null +++ b/model-00089-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad10dd048ca9e18e367f2213537dd825f7ebefca52d3203b5ead0dbf1d9ccfee +size 2095203852 diff --git a/model-00090-of-00189.safetensors b/model-00090-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9dcaa110fc8ad8539e9b202c953ba88be5a406d1 --- /dev/null +++ b/model-00090-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:932c530019f3f8edb7d88a97341efa0dbe5cf828a23e0209252c2b0b6588cc5a +size 1358954716 diff --git a/model-00091-of-00189.safetensors b/model-00091-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6e05eda20ee802f4c51f2cf34db0908a85eec82c --- /dev/null +++ b/model-00091-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f285e571f67533ec0881f1688fa2222c20a1b5dfd2522121fb225ce8197fd4f +size 2095203852 diff --git a/model-00092-of-00189.safetensors b/model-00092-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..14c231130c37b741e619792e89851813e3afc88a --- /dev/null +++ b/model-00092-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2fc0ec05e8fe34cc6f136b3055be36a2389895b135b9a4a26e020523f1235bf5 +size 1358954716 diff --git a/model-00093-of-00189.safetensors b/model-00093-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f6fce685f7d87620ac16dc0278e2175474019a81 --- /dev/null +++ b/model-00093-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef6e64ed3caacc011662d98a2b2b0c7c1ce2eef96848c7c7f673b265657f30fe +size 2095203852 diff --git a/model-00094-of-00189.safetensors b/model-00094-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e3757ae36cce13bed5e1ed2f247a94af865e05e0 --- /dev/null +++ b/model-00094-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:921c385deb74a2f2d3d7cad6b31b9ae6748faae43f10baa1bd079db5acfd0c05 +size 1358954716 diff --git a/model-00095-of-00189.safetensors b/model-00095-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cacc7d24adc6ecc0ea3242571889bf6c13c8f77e --- /dev/null +++ b/model-00095-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4dbf62137084e67331690ae9f4b701f7452cdea21786c7e765dd14322cbd1f76 +size 2095203852 diff --git a/model-00096-of-00189.safetensors b/model-00096-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8300f70e68550ff14ed825673fc96d0f1ab20229 --- /dev/null +++ b/model-00096-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:beac6084af7f90d6d8fe8637e69ab3c8bb7d4bceb47b7d0bdc4ac89420df8e77 +size 1358954716 diff --git a/model-00097-of-00189.safetensors b/model-00097-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2e52439bb1e45ec08958ef45ae9b6879145f2bdc --- /dev/null +++ b/model-00097-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12a0835e349405f32809002dc75421e1b875495e715f1d20fa42e4032012ce24 +size 2095203852 diff --git a/model-00098-of-00189.safetensors b/model-00098-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..29ab12ce9f943c8ed737391ff8a18d7a2dfc00bf --- /dev/null +++ b/model-00098-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d44bf53faf35d85ae867614f043e717f2a94eeee96a0f8c46e6e457327a21fad +size 1358954716 diff --git a/model-00099-of-00189.safetensors b/model-00099-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6bcfd033d76d6d5fed8c264fe15ceabe80ffd808 --- /dev/null +++ b/model-00099-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0649799fb34abddf7ac311a7b547bc0230bc32e8234b212b8f2d3a486d1f940e +size 2095203852 diff --git a/model-00100-of-00189.safetensors b/model-00100-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e8d42ee32f77ed8d6bcecb7e07ca71080e760515 --- /dev/null +++ b/model-00100-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b156516e3f912e3312bf66c5fe3297e770fe4376a0d237fa261e4465063b16d +size 1358954716 diff --git a/model-00101-of-00189.safetensors b/model-00101-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cfb4382a0921cdf89d09bd5568fec5e68f3f438c --- /dev/null +++ b/model-00101-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93f4a7bbccbb1e84c7465078e2cda32327cad9c502b9108a858368b84d153651 +size 2095203852 diff --git a/model-00102-of-00189.safetensors b/model-00102-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3ddd4f62f7ede6312e4eaaab6e21312b8633bad1 --- /dev/null +++ b/model-00102-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac803f4da0de817fc68088bd4c1b2e33e77e552f17b02ea880fae8916c51b7a1 +size 1358954716 diff --git a/model-00103-of-00189.safetensors b/model-00103-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..41a97b3ba817f163881e09abb11689a4c4be277b --- /dev/null +++ b/model-00103-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6fedd5621b627ae7f7cd6c6ac6cace0e2f19327504aab46b604689257231d61f +size 2095203852 diff --git a/model-00104-of-00189.safetensors b/model-00104-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..dcaa3ff8dfe35b66065cb5efd8591fc4f14dd648 --- /dev/null +++ b/model-00104-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86fb5638e09d944bcd6b378eaf36141fbb20d6906ed700f192ed936b37e4b95f +size 1358954716 diff --git a/model-00105-of-00189.safetensors b/model-00105-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b7c2cdaea7cb31f93aaed91051cfbf3daca30cc8 --- /dev/null +++ b/model-00105-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26aa232d28941d7766d0697a288b390fbfeb0653c6bd5e2b34ce985943624c1b +size 2095203852 diff --git a/model-00106-of-00189.safetensors b/model-00106-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fd721c6fa3478bbf854afd4a77c3465f0172dbfd --- /dev/null +++ b/model-00106-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54ab9ebe2fca796cfc142cb03be438b3af34756fd06dea099f121fb78da0bd6d +size 1358954716 diff --git a/model-00107-of-00189.safetensors b/model-00107-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0aa1b1773ec232b8f2daa67ef8b96e68ed0b2859 --- /dev/null +++ b/model-00107-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:813646a1b33c6a700395fded42e986daa235964ad6d2b70c49b9f78528a73486 +size 2095203852 diff --git a/model-00108-of-00189.safetensors b/model-00108-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..80811ede83786848deef4410adff2862189fb936 --- /dev/null +++ b/model-00108-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8fc04d4da86721516cd70587f9898a5ad1577295cdd810bfe8ad7f7868d385cb +size 1358954716 diff --git a/model-00109-of-00189.safetensors b/model-00109-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f54a7830eb57f0ed473f259d57ef1d304501d187 --- /dev/null +++ b/model-00109-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0bf05758233542d7efc699e4fe9d3d07325cb26141c842f53e2f4181f0e978a +size 2095203852 diff --git a/model-00110-of-00189.safetensors b/model-00110-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b43bd0b4557861237f37119ac5c0e3667a5bebb8 --- /dev/null +++ b/model-00110-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a23f47b63e6b7cf0aa6fe5adca8b27e5577d37151adc4504ae96a1f44a458c20 +size 1358954716 diff --git a/model-00111-of-00189.safetensors b/model-00111-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3f5494efc5cf48150c049e545cc16ee880c4a68e --- /dev/null +++ b/model-00111-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13fb34d400cff8d2e4dfd1c47c0b32dca030449875d511f3feeff4e64bd2ef97 +size 2095203852 diff --git a/model-00112-of-00189.safetensors b/model-00112-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..759250a11bc5ce63e665f25c5f7698309c4809e0 --- /dev/null +++ b/model-00112-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f319a812ca9216eb50fff13262a284da0601f54676f182499bc637ae98756c2 +size 1358954716 diff --git a/model-00113-of-00189.safetensors b/model-00113-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6b8be4c5f10e087fb0aa575e29b36cfe3ccdb04c --- /dev/null +++ b/model-00113-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf7f3e549ff1b413d139ca9bd5798c25c8b61200e4e0e00e329f544b5013eab0 +size 2095203852 diff --git a/model-00114-of-00189.safetensors b/model-00114-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f3f7324616b64dc06e42cbc6e7fe3ebae324a734 --- /dev/null +++ b/model-00114-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c35c84063db877312718556656cf4fd66b58fb8829f52710cd9cce00a1cd134 +size 1358954716 diff --git a/model-00115-of-00189.safetensors b/model-00115-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..52c0d07f33658030a1435802ef68000db6748ac4 --- /dev/null +++ b/model-00115-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7fe845d724b15c2a8a87fa18c811d38e0ce7cbf9f23d1afd449b95e977ad5fd0 +size 2095203852 diff --git a/model-00116-of-00189.safetensors b/model-00116-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..319532913e9d7467c487f9721c5ce5f2e6cd6659 --- /dev/null +++ b/model-00116-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:305d45b28f7493d45f2a96b829d468a884b3ea64796549901d4912c49dbf7a64 +size 1358954716 diff --git a/model-00117-of-00189.safetensors b/model-00117-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6ce6f7589247b17f25d697858241831a75eb6a1e --- /dev/null +++ b/model-00117-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d650951014597a443b8226259ba798eea865a10b858a3590c9705478a686fcca +size 2095203852 diff --git a/model-00118-of-00189.safetensors b/model-00118-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..69919ed46b90ebc4c05a09531e768071e1492ce8 --- /dev/null +++ b/model-00118-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4052b9ab2b1e17884190dd75cf1688370931238cd62557e144cf1407890ff97 +size 1358954716 diff --git a/model-00119-of-00189.safetensors b/model-00119-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5c2fc336f358794017439876cf608761c588848a --- /dev/null +++ b/model-00119-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c55db5f11ac134df917a8c37dd34bd76bfbda8f3d10db124ea8b7929b29db20 +size 2095203852 diff --git a/model-00120-of-00189.safetensors b/model-00120-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e37f2c3e92c3a7dbf55d48d631f3ecb0d620d829 --- /dev/null +++ b/model-00120-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb0f259b0524c2a4cd4e06f3b07eb10d67d1deee555224d5ce537b616d11a239 +size 1358954716 diff --git a/model-00121-of-00189.safetensors b/model-00121-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..796ac50832ad6594ed45ae367a60434777057b63 --- /dev/null +++ b/model-00121-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b762553bda6cb75b5e7a96f4a13d9378974b251264e5d93fbe560ad5339227e9 +size 2095203852 diff --git a/model-00122-of-00189.safetensors b/model-00122-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f08c72a9dc3ad70e0ce4ad03f1392d34a5d87d86 --- /dev/null +++ b/model-00122-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f981b6a03cb149b8106f9362482e0bbcd3c7a12ceaf4dd1790f2a200b1aee5f5 +size 1358954716 diff --git a/model-00123-of-00189.safetensors b/model-00123-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..339e87133d5d2d15fecca10b682bdf4123168ee1 --- /dev/null +++ b/model-00123-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d81915f864fe8cf04170c814588fd57ab3b9477570a2bca98b83a2668388bb6b +size 2095203852 diff --git a/model-00124-of-00189.safetensors b/model-00124-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6fa4c749d4f25c76fa7862fc7604403a62b2882f --- /dev/null +++ b/model-00124-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e99b3f7ca1dcf36960fcfa17dd57b0a6955c967d40b8ccd151a3c28eb6eeea1 +size 1358954716 diff --git a/model-00125-of-00189.safetensors b/model-00125-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5cc63da8eb1252396bd679fdcdf6550f6224d840 --- /dev/null +++ b/model-00125-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fdc30170783ae35ac665f49d8e0415a610135947a75b632f103e61976180d5ec +size 2095203852 diff --git a/model-00126-of-00189.safetensors b/model-00126-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e58612720c6593a2560152022898c03dc8f9c87e --- /dev/null +++ b/model-00126-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2bafb0ad1252d1fb4b355ecf16c5475fe157e091ab40b0dc9accecadc1ed352 +size 1358954716 diff --git a/model-00127-of-00189.safetensors b/model-00127-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f0de175fb669781ef01331b6ef84000d883dfe8d --- /dev/null +++ b/model-00127-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67cea4b00b9e2c8627e77709bf8df1fb4e326b22aa4acbf37ef12b4f59738370 +size 2095203852 diff --git a/model-00128-of-00189.safetensors b/model-00128-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..798968717520193151fb5be3187bf7d37beb1a39 --- /dev/null +++ b/model-00128-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9096c6345dd60353e351e888ce1d791e14b536430083133197fa23c919cd4ee0 +size 1358954716 diff --git a/model-00129-of-00189.safetensors b/model-00129-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..10de87f7f0f5d32013a698146c7b621023551349 --- /dev/null +++ b/model-00129-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:633784256b1d770de93dd2701cd5d48c738fdd23e6c43434f6cb9dd6ca020813 +size 2095203852 diff --git a/model-00130-of-00189.safetensors b/model-00130-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..08087d5718ca4cdcd65d5bd3eb2b6f1745d8c2a7 --- /dev/null +++ b/model-00130-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53196cd79577730e6d466d27aeb6c455237b416ef59bc03a78703076314abe5a +size 1358954716 diff --git a/model-00131-of-00189.safetensors b/model-00131-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..97ab6ded93d693e2972484b3f21114bdc091f92a --- /dev/null +++ b/model-00131-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a4b714b44c50c26899eea277bb22ae41470d568ed23d172af13362faafdc2c6 +size 2095203852 diff --git a/model-00132-of-00189.safetensors b/model-00132-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d00124dbc2bcad4d4fac0d5810d5affc2806d15e --- /dev/null +++ b/model-00132-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80334cefd9583cbbe57f2f5dedc5e1ee0c4d459d7682e938607ab4bf83a7f968 +size 1358954716 diff --git a/model-00133-of-00189.safetensors b/model-00133-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..60561c5e7f54b136037120004f50b4353b8e49c9 --- /dev/null +++ b/model-00133-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1742d703ed03d92ea2b7838d07f90cf0efdf44e3b36c0fd13a7bf7a01e8f2a00 +size 2095203852 diff --git a/model-00134-of-00189.safetensors b/model-00134-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..294c90660648eb44985bd7cb01a50f6d7857c3d2 --- /dev/null +++ b/model-00134-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61b0672df1495a8ae3b2b8576cbc9cead683cbf508120cc7723bf1b17cb5d5b8 +size 1358954716 diff --git a/model-00135-of-00189.safetensors b/model-00135-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d378981f0fb50160d0ac5aaf3fb5b7d9ba7c781e --- /dev/null +++ b/model-00135-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a83a92f0a2d05bb0ba6de9a2fb59529061fee17378d57ba3dcf1d6e200e1c71 +size 2095203852 diff --git a/model-00136-of-00189.safetensors b/model-00136-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..478b4f6f41e58f808145b1080447f54f17c08f6c --- /dev/null +++ b/model-00136-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c5373c97d6e619b5b30aea913d392f55a876a5b0ec64897a75ef62d1a4a3eb0 +size 1358954716 diff --git a/model-00137-of-00189.safetensors b/model-00137-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0577494990693d07894bda48a8215e9b5c828c5c --- /dev/null +++ b/model-00137-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:232e32e55611d675207dc6087a1745fe49ee438ad1b23edf355b9913cc67dc78 +size 2095203852 diff --git a/model-00138-of-00189.safetensors b/model-00138-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a036725d2f97caedce1d9fabdb8d6117a1a1afc1 --- /dev/null +++ b/model-00138-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41061eba2f09330e66abc490725eacb198c06c88197cfbecc46f5a13789e8a2f +size 1358954716 diff --git a/model-00139-of-00189.safetensors b/model-00139-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3e5ec9854f2b6c2d4ea4d4066c13eaa4b1249170 --- /dev/null +++ b/model-00139-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73fa925c40634360a4e82c0bdec73d36a7b4b6bae19373ec4351fb9ff4ab8d43 +size 2095203852 diff --git a/model-00140-of-00189.safetensors b/model-00140-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..75282555d7a381b951b4b7214d5db3baa0557b44 --- /dev/null +++ b/model-00140-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d559d16bda38d7ccd2826bec1b4a11cf0eb2dcb65231ff5f4d3ba3ea99bf186d +size 1358954716 diff --git a/model-00141-of-00189.safetensors b/model-00141-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f8ce46fa2ce84f19daa401eb2f0750d252b31b30 --- /dev/null +++ b/model-00141-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18071946a030ee05d4a2ba2d1051f44488f899414c93f1c0e3cde202a8c0a1f7 +size 2095203852 diff --git a/model-00142-of-00189.safetensors b/model-00142-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cfb8d6694bd545d6a2b3eed06c300f36f179673a --- /dev/null +++ b/model-00142-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a289e22cf06a72d886e09e53ca87772685825715f35b7e5954da24fb939d3fb9 +size 1358954716 diff --git a/model-00143-of-00189.safetensors b/model-00143-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ed70be7794144f77c27c1eea888c19cc16653f8e --- /dev/null +++ b/model-00143-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d1c7868172c9687b7c5fb50cf8b5818b6eb849767936bb1dbbce612f02e46ba +size 2095203852 diff --git a/model-00144-of-00189.safetensors b/model-00144-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1731b2c64611553360cdf1bbe21fc9e9c6276134 --- /dev/null +++ b/model-00144-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8465946019d18a7ee9461d8381e16f2030244c9e51a6c1dcf3815ce6b7aacce1 +size 1358954716 diff --git a/model-00145-of-00189.safetensors b/model-00145-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..57b3c408ce97d9b081ad17c380c5f3fe46fcae70 --- /dev/null +++ b/model-00145-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75d659c524c31914670bc29619fe1399a8dad63a8a3b3437ebb9dc050fc439b9 +size 2095203852 diff --git a/model-00146-of-00189.safetensors b/model-00146-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..dc4f7b85885339f2f25fc179c70d18407725fc5d --- /dev/null +++ b/model-00146-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3162b1bf5252f782a9f02b4aa1920acbadaa6ed98262f158b8fc53f5e6882039 +size 1358954716 diff --git a/model-00147-of-00189.safetensors b/model-00147-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e62cc8fda9f95bcf926b0c36828e24ad9602fb57 --- /dev/null +++ b/model-00147-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3fa4cda61f58d9e61a0ab38c50719b1f2bc6607be25d540833ed5a4221d28aa +size 2095203852 diff --git a/model-00148-of-00189.safetensors b/model-00148-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..dbf48c722cda5815b1d778761dd10e954f5d57b8 --- /dev/null +++ b/model-00148-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3d1ed9a18ff6198e7abe42676b58644195e86e09a702ea6c6678786e9e93788 +size 1358954716 diff --git a/model-00149-of-00189.safetensors b/model-00149-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cdcf9fabee6237837dbdcd676d50e5bc71e09059 --- /dev/null +++ b/model-00149-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:533aa7006672d297ef413f766cf9239df098a518fd145b65b3a8c76ebfd565ac +size 2095203852 diff --git a/model-00150-of-00189.safetensors b/model-00150-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6a56544edbbc47cd8d2d4ef6800f1d9f20b43096 --- /dev/null +++ b/model-00150-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20e2cd3031e9780e7019d32a7a8d3dbcedbcd2c2bb3dcb40bbfc3bc9ead5d9d4 +size 1358954716 diff --git a/model-00151-of-00189.safetensors b/model-00151-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2358748d58a793caae1eae5fb0a27df7be639b2d --- /dev/null +++ b/model-00151-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e42605ba7db8cf15e8caac9fc62133b21bfa54c91f521171e1def26f779d043e +size 2095203852 diff --git a/model-00152-of-00189.safetensors b/model-00152-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..5ba664b8cc1722472a0bb41bf171de575342fc58 --- /dev/null +++ b/model-00152-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:376798644f119c486e6fe32bafdad96069a27fe1e1023ae7ce0ce07591238d94 +size 1358954716 diff --git a/model-00153-of-00189.safetensors b/model-00153-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c248665406d00f5b569552274a8f53f3fc8fc103 --- /dev/null +++ b/model-00153-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48444bad76bff2638f5129645c101cd6053ac3fd77c4f2148b9366846f5fc02c +size 2095203852 diff --git a/model-00154-of-00189.safetensors b/model-00154-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6c3a5c8978010f96e0860fa0852fa2c3afe7a220 --- /dev/null +++ b/model-00154-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f8e00702daa7cc17cd4d532bf1c1eeba90ca78084096b2aabc46b72952ad079b +size 1358954716 diff --git a/model-00155-of-00189.safetensors b/model-00155-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c5b9c6dd16662ef02bc5fcc71d0a1ac3413ae3d1 --- /dev/null +++ b/model-00155-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0fda17dd9f699d2de058d223a57b7902ffb8150f1ab725c6de91b2fae619684 +size 2095203852 diff --git a/model-00156-of-00189.safetensors b/model-00156-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..75c871eba10d377ec68a66db65d115aa7c2711f9 --- /dev/null +++ b/model-00156-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7008809dcafd38aabc57a2eb9a6b27dfc4c0c3ef3a39818eda12afe3b3228f0f +size 1358954716 diff --git a/model-00157-of-00189.safetensors b/model-00157-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..81cbb24abc2862bbbce0f21c9b8e39fa418888f7 --- /dev/null +++ b/model-00157-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1fba0f9e24b3511c2fe1f846b56c1cfa15106a13c7dfd199cfce17880d9b052b +size 2095203852 diff --git a/model-00158-of-00189.safetensors b/model-00158-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9429181ea8994ac700b9e5f83509c7f1750aa5e3 --- /dev/null +++ b/model-00158-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81da71c1c6f842e9a3c41e909f6c94e5082d385b3716935e21b976302eb9f4e6 +size 1358954716 diff --git a/model-00159-of-00189.safetensors b/model-00159-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..46efdacc91f26532d739c162f112c8ab7fd5bf7c --- /dev/null +++ b/model-00159-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ec8104ff7f82c9ec406f0c78c0e414bb612f4a492ccb7fa6f7721c0b496b24a +size 2095203852 diff --git a/model-00160-of-00189.safetensors b/model-00160-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..735af9ae0acaa9d745da3128590a06187989ef1c --- /dev/null +++ b/model-00160-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:783679058605ce96f61d04263a01c7256a3b2c9568c5345302f92d96d7bd551a +size 1358954716 diff --git a/model-00161-of-00189.safetensors b/model-00161-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..03ff9c081c4aecee7e999619c86f75f93cb4a3e4 --- /dev/null +++ b/model-00161-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fef3214ca452c3dc107f7adc6f5f9095b0b28f52fa3283163505e2d72d29499d +size 2095203852 diff --git a/model-00162-of-00189.safetensors b/model-00162-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3c0ee7f6f5730e036cd04791ab6ab8ec4d6332a6 --- /dev/null +++ b/model-00162-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a152153ae193deea2eb3a85c6929d8c1f074e012fc8afff8359afcdde81be967 +size 1358954716 diff --git a/model-00163-of-00189.safetensors b/model-00163-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ecacff2368bb7ad7777087bdb2c52d285fb67678 --- /dev/null +++ b/model-00163-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16a4a02c2388558f37387d5e6745193b07357d4206ebcc2bb8a2ba369850edc8 +size 2095203852 diff --git a/model-00164-of-00189.safetensors b/model-00164-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f50a2c5b3b41714193d2fd840a56b5e058373a54 --- /dev/null +++ b/model-00164-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9fd569956a395d2ccb47433cca20a6209375a7d69b5c8ed36351bb7f53d04ae +size 1358954716 diff --git a/model-00165-of-00189.safetensors b/model-00165-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..53e4150f4eeda61a2ad2db98698a3bf83f9fcd85 --- /dev/null +++ b/model-00165-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db08be6ce8351c50fb0993622c208de1eb8fe25bf0e90024cea9240cba5b2c83 +size 2095203852 diff --git a/model-00166-of-00189.safetensors b/model-00166-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b985391da218e8f3ab16b97e7c379b59394f9580 --- /dev/null +++ b/model-00166-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f763e93aa4944c563d4e4b679d3969264fb3ed6ddc931ae6351e3d3a66c0bf0 +size 1358954716 diff --git a/model-00167-of-00189.safetensors b/model-00167-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f5c9f6868e37568a1b666558e14705181104ff42 --- /dev/null +++ b/model-00167-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:040f11db4d37fd12262a77d64818ae129c2b85eac002540c1a7ff779334fb676 +size 2095203852 diff --git a/model-00168-of-00189.safetensors b/model-00168-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..12105417c1f88e83b48ee34a99a07b9de2ee861a --- /dev/null +++ b/model-00168-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fb57dab8bab64e72fbdfd39675ea3231a0dc83ef84b96c0db11b1374dcd13090 +size 1358954716 diff --git a/model-00169-of-00189.safetensors b/model-00169-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fb72aad78532137b166af5ae0a8fc9f60ae713b5 --- /dev/null +++ b/model-00169-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa9c1bc0e5a563317eb4a03345befdf4ed6af59afd2e82a0851cdb8221b1ff4b +size 2095203852 diff --git a/model-00170-of-00189.safetensors b/model-00170-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..416c34eae333ef2e201822210b2fd020781ff5fb --- /dev/null +++ b/model-00170-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8abf3659a01be23e827e23aef2a79d15357b530cd10160e545775f404849b7c3 +size 1358954716 diff --git a/model-00171-of-00189.safetensors b/model-00171-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f7f7a09f8b746ab9d2542b96f760c9d1a1780df7 --- /dev/null +++ b/model-00171-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19f218b1d46512cbff59042a71e755354ef7edc0c42e046830467cce5de83a2a +size 2095203852 diff --git a/model-00172-of-00189.safetensors b/model-00172-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..de13b1b371a14c8f58bf00f08c2847ce75d9a9d1 --- /dev/null +++ b/model-00172-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d41d75f2f8215967be331d1ac45204f2298e0dd911697a0571d67509deb373ad +size 1358954716 diff --git a/model-00173-of-00189.safetensors b/model-00173-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..67fd3bd3052c91ca5a3349b367734d255272c62e --- /dev/null +++ b/model-00173-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0dea77ca5ca9a3c1c61ae1ad26de4fc74e2101605285e226061a3df0629f8323 +size 2095203852 diff --git a/model-00174-of-00189.safetensors b/model-00174-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..de758669736f889bc8885b31699069cf30425120 --- /dev/null +++ b/model-00174-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:015ccc05f8eda2920737d5cfddad947dc9cef1a4df3660d0f93fae790c01d4f7 +size 1358954716 diff --git a/model-00175-of-00189.safetensors b/model-00175-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..946f07fd12d5685f9837890c7e0e0cf49cfa201c --- /dev/null +++ b/model-00175-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f432fadd6860edae5cbb2441e9f113f15da9e9ca7bcf8fa63ae59df8c381b8ce +size 2095203852 diff --git a/model-00176-of-00189.safetensors b/model-00176-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9336737b5bc2e0f56e96a60e5be13c02324b65d7 --- /dev/null +++ b/model-00176-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4836f40e450da004be4ce561ab8a673aa29876e79d78cab4eb8b8a2fa50f60b6 +size 1358954716 diff --git a/model-00177-of-00189.safetensors b/model-00177-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2731e31c221ea26d84f5ecba998d916a92922895 --- /dev/null +++ b/model-00177-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc069f0f768807d92878fd421c3c90a25c424ac7117bc7950d8d61c04ab0210f +size 2095203852 diff --git a/model-00178-of-00189.safetensors b/model-00178-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9f448a4d91ae0ba3022b9468e0563685841cba2c --- /dev/null +++ b/model-00178-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1efe468594197d081c1be9e82645f8a4d2f6dc137d8ac41568e6f501a4c086fc +size 1358954716 diff --git a/model-00179-of-00189.safetensors b/model-00179-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a3041a903733713ef68463ca216f0a39f4552208 --- /dev/null +++ b/model-00179-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f16f48d8d9bc5694c87500a4e9304f98482641b772f57cb3b022ff7b3510982d +size 2095203852 diff --git a/model-00180-of-00189.safetensors b/model-00180-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3c3eef095f27cb589ee7a824d2947e1af2effd93 --- /dev/null +++ b/model-00180-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8241e46d07aa5eda5c767bb6485e1c53d08fe258a7ccc8d8adc84caf54ee7e6 +size 1358954716 diff --git a/model-00181-of-00189.safetensors b/model-00181-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..37e81225b2b20a63d438c0cab761ba9c325beae4 --- /dev/null +++ b/model-00181-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bcf9e4154733102d597382fe159732b65c7ff5de7dd7f9051a23be4c5186c829 +size 2095203852 diff --git a/model-00182-of-00189.safetensors b/model-00182-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..286acc51d7fe3c3ddd4d020d60677210c2a3d266 --- /dev/null +++ b/model-00182-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ec104ee73ec17ccd2eb17616c2bdc48c17e56c3030f12113af766837d4222fe +size 1358954716 diff --git a/model-00183-of-00189.safetensors b/model-00183-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..779c5a1750400176033cbba8f5222a75a9f32d2f --- /dev/null +++ b/model-00183-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c86c4d0b079571dab2ba9344e071de31e6834e65063012b0dc80e2b7a951bac +size 2095203852 diff --git a/model-00184-of-00189.safetensors b/model-00184-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0208b9cc743e4567f02851dba2efa0a0c5663455 --- /dev/null +++ b/model-00184-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64ecf862f409f78f97c5b07cae7173ccec4faadd4e25b3d45c93868ccfbea39f +size 1358954716 diff --git a/model-00185-of-00189.safetensors b/model-00185-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bd19a152828be7becb26b3d6bf37bae33d159bd0 --- /dev/null +++ b/model-00185-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e08852165aa57f6b1358adeb36b2dee60183acd5d55f919758cf275363c6e57 +size 2095203852 diff --git a/model-00186-of-00189.safetensors b/model-00186-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bb4d860e2cc8147bdd25b53c09fcbebb6c9deee7 --- /dev/null +++ b/model-00186-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:addcbb4bf5fe6409af2cf62d2147cd6bd36250db7789c26372588d42bd31760a +size 1358954716 diff --git a/model-00187-of-00189.safetensors b/model-00187-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6e5b437a59efa5689edbe6b20cb9beff854badf8 --- /dev/null +++ b/model-00187-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d29046a593cea78f17e88750c7028d73dcb101670c661bb22b5ab3382113d5b8 +size 3567406348 diff --git a/model-00188-of-00189.safetensors b/model-00188-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..79b6f98a19c1f58e45e177c438527aaf5f7c958a --- /dev/null +++ b/model-00188-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7bdb5dfd148e432772a4755439b2f45cb9308dbdc2105143c2565ee3f74ef36f +size 9437184096 diff --git a/model-00189-of-00189.safetensors b/model-00189-of-00189.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6027003801775ffd4f01bf86a7f46d27e24c6c91 --- /dev/null +++ b/model-00189-of-00189.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4aebf940dc80a3d84ed9f8b6d22635bb7500d5739443066698defb58ac4f8273 +size 8154245840 diff --git a/model.safetensors.index.json b/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..945660c1da523300d1142967dfc94e25c12c9868 --- /dev/null +++ b/model.safetensors.index.json @@ -0,0 +1,1547 @@ +{ + "metadata": { + "total_size": 350473664368 + }, + "weight_map": { + "lm_head.weight": "model-00188-of-00189.safetensors", + "model.embed_tokens.weight": "model-00001-of-00189.safetensors", + "model.layers.0.input_layernorm.bias": "model-00003-of-00189.safetensors", + "model.layers.0.input_layernorm.weight": "model-00003-of-00189.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00003-of-00189.safetensors", + "model.layers.0.mlp.down_proj.weight_scale": "model-00003-of-00189.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00002-of-00189.safetensors", + "model.layers.0.mlp.up_proj.weight_scale": "model-00002-of-00189.safetensors", + "model.layers.0.post_attention_layernorm.bias": "model-00003-of-00189.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00003-of-00189.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00002-of-00189.safetensors", + "model.layers.0.self_attn.k_proj.weight_scale": "model-00002-of-00189.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00002-of-00189.safetensors", + "model.layers.0.self_attn.o_proj.weight_scale": "model-00002-of-00189.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00002-of-00189.safetensors", + "model.layers.0.self_attn.q_proj.weight_scale": "model-00002-of-00189.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00002-of-00189.safetensors", + "model.layers.0.self_attn.v_proj.weight_scale": "model-00002-of-00189.safetensors", + "model.layers.1.input_layernorm.bias": "model-00005-of-00189.safetensors", + "model.layers.1.input_layernorm.weight": "model-00005-of-00189.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00005-of-00189.safetensors", + "model.layers.1.mlp.down_proj.weight_scale": "model-00005-of-00189.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00004-of-00189.safetensors", + "model.layers.1.mlp.up_proj.weight_scale": "model-00004-of-00189.safetensors", + "model.layers.1.post_attention_layernorm.bias": "model-00005-of-00189.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00005-of-00189.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00003-of-00189.safetensors", + "model.layers.1.self_attn.k_proj.weight_scale": "model-00003-of-00189.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00003-of-00189.safetensors", + "model.layers.1.self_attn.o_proj.weight_scale": "model-00003-of-00189.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00003-of-00189.safetensors", + "model.layers.1.self_attn.q_proj.weight_scale": "model-00003-of-00189.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00003-of-00189.safetensors", + "model.layers.1.self_attn.v_proj.weight_scale": "model-00003-of-00189.safetensors", + "model.layers.10.input_layernorm.bias": "model-00023-of-00189.safetensors", + "model.layers.10.input_layernorm.weight": "model-00023-of-00189.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00023-of-00189.safetensors", + "model.layers.10.mlp.down_proj.weight_scale": "model-00023-of-00189.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00022-of-00189.safetensors", + "model.layers.10.mlp.up_proj.weight_scale": "model-00022-of-00189.safetensors", + "model.layers.10.post_attention_layernorm.bias": "model-00023-of-00189.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00023-of-00189.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00021-of-00189.safetensors", + "model.layers.10.self_attn.k_proj.weight_scale": "model-00021-of-00189.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00021-of-00189.safetensors", + "model.layers.10.self_attn.o_proj.weight_scale": "model-00021-of-00189.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00021-of-00189.safetensors", + "model.layers.10.self_attn.q_proj.weight_scale": "model-00021-of-00189.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00021-of-00189.safetensors", + "model.layers.10.self_attn.v_proj.weight_scale": "model-00021-of-00189.safetensors", + "model.layers.11.input_layernorm.bias": "model-00025-of-00189.safetensors", + "model.layers.11.input_layernorm.weight": "model-00025-of-00189.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00025-of-00189.safetensors", + "model.layers.11.mlp.down_proj.weight_scale": "model-00025-of-00189.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00024-of-00189.safetensors", + "model.layers.11.mlp.up_proj.weight_scale": "model-00024-of-00189.safetensors", + "model.layers.11.post_attention_layernorm.bias": "model-00025-of-00189.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00025-of-00189.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00023-of-00189.safetensors", + "model.layers.11.self_attn.k_proj.weight_scale": "model-00023-of-00189.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00023-of-00189.safetensors", + "model.layers.11.self_attn.o_proj.weight_scale": "model-00023-of-00189.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00023-of-00189.safetensors", + "model.layers.11.self_attn.q_proj.weight_scale": "model-00023-of-00189.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00023-of-00189.safetensors", + "model.layers.11.self_attn.v_proj.weight_scale": "model-00023-of-00189.safetensors", + "model.layers.12.input_layernorm.bias": "model-00027-of-00189.safetensors", + "model.layers.12.input_layernorm.weight": "model-00027-of-00189.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00027-of-00189.safetensors", + "model.layers.12.mlp.down_proj.weight_scale": "model-00027-of-00189.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00026-of-00189.safetensors", + "model.layers.12.mlp.up_proj.weight_scale": "model-00026-of-00189.safetensors", + "model.layers.12.post_attention_layernorm.bias": "model-00027-of-00189.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00027-of-00189.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00025-of-00189.safetensors", + "model.layers.12.self_attn.k_proj.weight_scale": "model-00025-of-00189.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00025-of-00189.safetensors", + "model.layers.12.self_attn.o_proj.weight_scale": "model-00025-of-00189.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00025-of-00189.safetensors", + "model.layers.12.self_attn.q_proj.weight_scale": "model-00025-of-00189.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00025-of-00189.safetensors", + "model.layers.12.self_attn.v_proj.weight_scale": "model-00025-of-00189.safetensors", + "model.layers.13.input_layernorm.bias": "model-00029-of-00189.safetensors", + "model.layers.13.input_layernorm.weight": "model-00029-of-00189.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00029-of-00189.safetensors", + "model.layers.13.mlp.down_proj.weight_scale": "model-00029-of-00189.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00028-of-00189.safetensors", + "model.layers.13.mlp.up_proj.weight_scale": "model-00028-of-00189.safetensors", + "model.layers.13.post_attention_layernorm.bias": "model-00029-of-00189.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00029-of-00189.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00027-of-00189.safetensors", + "model.layers.13.self_attn.k_proj.weight_scale": "model-00027-of-00189.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00027-of-00189.safetensors", + "model.layers.13.self_attn.o_proj.weight_scale": "model-00027-of-00189.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00027-of-00189.safetensors", + "model.layers.13.self_attn.q_proj.weight_scale": "model-00027-of-00189.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00027-of-00189.safetensors", + "model.layers.13.self_attn.v_proj.weight_scale": "model-00027-of-00189.safetensors", + "model.layers.14.input_layernorm.bias": "model-00031-of-00189.safetensors", + "model.layers.14.input_layernorm.weight": "model-00031-of-00189.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00031-of-00189.safetensors", + "model.layers.14.mlp.down_proj.weight_scale": "model-00031-of-00189.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00030-of-00189.safetensors", + "model.layers.14.mlp.up_proj.weight_scale": "model-00030-of-00189.safetensors", + "model.layers.14.post_attention_layernorm.bias": "model-00031-of-00189.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00031-of-00189.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00029-of-00189.safetensors", + "model.layers.14.self_attn.k_proj.weight_scale": "model-00029-of-00189.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00029-of-00189.safetensors", + "model.layers.14.self_attn.o_proj.weight_scale": "model-00029-of-00189.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00029-of-00189.safetensors", + "model.layers.14.self_attn.q_proj.weight_scale": "model-00029-of-00189.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00029-of-00189.safetensors", + "model.layers.14.self_attn.v_proj.weight_scale": "model-00029-of-00189.safetensors", + "model.layers.15.input_layernorm.bias": "model-00033-of-00189.safetensors", + "model.layers.15.input_layernorm.weight": "model-00033-of-00189.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00033-of-00189.safetensors", + "model.layers.15.mlp.down_proj.weight_scale": "model-00033-of-00189.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00032-of-00189.safetensors", + "model.layers.15.mlp.up_proj.weight_scale": "model-00032-of-00189.safetensors", + "model.layers.15.post_attention_layernorm.bias": "model-00033-of-00189.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00033-of-00189.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00031-of-00189.safetensors", + "model.layers.15.self_attn.k_proj.weight_scale": "model-00031-of-00189.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00031-of-00189.safetensors", + "model.layers.15.self_attn.o_proj.weight_scale": "model-00031-of-00189.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00031-of-00189.safetensors", + "model.layers.15.self_attn.q_proj.weight_scale": "model-00031-of-00189.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00031-of-00189.safetensors", + "model.layers.15.self_attn.v_proj.weight_scale": "model-00031-of-00189.safetensors", + "model.layers.16.input_layernorm.bias": "model-00035-of-00189.safetensors", + "model.layers.16.input_layernorm.weight": "model-00035-of-00189.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00035-of-00189.safetensors", + "model.layers.16.mlp.down_proj.weight_scale": "model-00035-of-00189.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00034-of-00189.safetensors", + "model.layers.16.mlp.up_proj.weight_scale": "model-00034-of-00189.safetensors", + "model.layers.16.post_attention_layernorm.bias": "model-00035-of-00189.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00035-of-00189.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00033-of-00189.safetensors", + "model.layers.16.self_attn.k_proj.weight_scale": "model-00033-of-00189.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00033-of-00189.safetensors", + "model.layers.16.self_attn.o_proj.weight_scale": "model-00033-of-00189.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00033-of-00189.safetensors", + "model.layers.16.self_attn.q_proj.weight_scale": "model-00033-of-00189.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00033-of-00189.safetensors", + "model.layers.16.self_attn.v_proj.weight_scale": "model-00033-of-00189.safetensors", + "model.layers.17.input_layernorm.bias": "model-00037-of-00189.safetensors", + "model.layers.17.input_layernorm.weight": "model-00037-of-00189.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00037-of-00189.safetensors", + "model.layers.17.mlp.down_proj.weight_scale": "model-00037-of-00189.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00036-of-00189.safetensors", + "model.layers.17.mlp.up_proj.weight_scale": "model-00036-of-00189.safetensors", + "model.layers.17.post_attention_layernorm.bias": "model-00037-of-00189.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00037-of-00189.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00035-of-00189.safetensors", + "model.layers.17.self_attn.k_proj.weight_scale": "model-00035-of-00189.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00035-of-00189.safetensors", + "model.layers.17.self_attn.o_proj.weight_scale": "model-00035-of-00189.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00035-of-00189.safetensors", + "model.layers.17.self_attn.q_proj.weight_scale": "model-00035-of-00189.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00035-of-00189.safetensors", + "model.layers.17.self_attn.v_proj.weight_scale": "model-00035-of-00189.safetensors", + "model.layers.18.input_layernorm.bias": "model-00039-of-00189.safetensors", + "model.layers.18.input_layernorm.weight": "model-00039-of-00189.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00039-of-00189.safetensors", + "model.layers.18.mlp.down_proj.weight_scale": "model-00039-of-00189.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00038-of-00189.safetensors", + "model.layers.18.mlp.up_proj.weight_scale": "model-00038-of-00189.safetensors", + "model.layers.18.post_attention_layernorm.bias": "model-00039-of-00189.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00039-of-00189.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00037-of-00189.safetensors", + "model.layers.18.self_attn.k_proj.weight_scale": "model-00037-of-00189.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00037-of-00189.safetensors", + "model.layers.18.self_attn.o_proj.weight_scale": "model-00037-of-00189.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00037-of-00189.safetensors", + "model.layers.18.self_attn.q_proj.weight_scale": "model-00037-of-00189.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00037-of-00189.safetensors", + "model.layers.18.self_attn.v_proj.weight_scale": "model-00037-of-00189.safetensors", + "model.layers.19.input_layernorm.bias": "model-00041-of-00189.safetensors", + "model.layers.19.input_layernorm.weight": "model-00041-of-00189.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00041-of-00189.safetensors", + "model.layers.19.mlp.down_proj.weight_scale": "model-00041-of-00189.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00040-of-00189.safetensors", + "model.layers.19.mlp.up_proj.weight_scale": "model-00040-of-00189.safetensors", + "model.layers.19.post_attention_layernorm.bias": "model-00041-of-00189.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00041-of-00189.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00039-of-00189.safetensors", + "model.layers.19.self_attn.k_proj.weight_scale": "model-00039-of-00189.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00039-of-00189.safetensors", + "model.layers.19.self_attn.o_proj.weight_scale": "model-00039-of-00189.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00039-of-00189.safetensors", + "model.layers.19.self_attn.q_proj.weight_scale": "model-00039-of-00189.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00039-of-00189.safetensors", + "model.layers.19.self_attn.v_proj.weight_scale": "model-00039-of-00189.safetensors", + "model.layers.2.input_layernorm.bias": "model-00007-of-00189.safetensors", + "model.layers.2.input_layernorm.weight": "model-00007-of-00189.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00007-of-00189.safetensors", + "model.layers.2.mlp.down_proj.weight_scale": "model-00007-of-00189.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00006-of-00189.safetensors", + "model.layers.2.mlp.up_proj.weight_scale": "model-00006-of-00189.safetensors", + "model.layers.2.post_attention_layernorm.bias": "model-00007-of-00189.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00007-of-00189.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00005-of-00189.safetensors", + "model.layers.2.self_attn.k_proj.weight_scale": "model-00005-of-00189.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00005-of-00189.safetensors", + "model.layers.2.self_attn.o_proj.weight_scale": "model-00005-of-00189.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00005-of-00189.safetensors", + "model.layers.2.self_attn.q_proj.weight_scale": "model-00005-of-00189.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00005-of-00189.safetensors", + "model.layers.2.self_attn.v_proj.weight_scale": "model-00005-of-00189.safetensors", + "model.layers.20.input_layernorm.bias": "model-00043-of-00189.safetensors", + "model.layers.20.input_layernorm.weight": "model-00043-of-00189.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00043-of-00189.safetensors", + "model.layers.20.mlp.down_proj.weight_scale": "model-00043-of-00189.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00042-of-00189.safetensors", + "model.layers.20.mlp.up_proj.weight_scale": "model-00042-of-00189.safetensors", + "model.layers.20.post_attention_layernorm.bias": "model-00043-of-00189.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00043-of-00189.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00041-of-00189.safetensors", + "model.layers.20.self_attn.k_proj.weight_scale": "model-00041-of-00189.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00041-of-00189.safetensors", + "model.layers.20.self_attn.o_proj.weight_scale": "model-00041-of-00189.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00041-of-00189.safetensors", + "model.layers.20.self_attn.q_proj.weight_scale": "model-00041-of-00189.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00041-of-00189.safetensors", + "model.layers.20.self_attn.v_proj.weight_scale": "model-00041-of-00189.safetensors", + "model.layers.21.input_layernorm.bias": "model-00045-of-00189.safetensors", + "model.layers.21.input_layernorm.weight": "model-00045-of-00189.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00045-of-00189.safetensors", + "model.layers.21.mlp.down_proj.weight_scale": "model-00045-of-00189.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00044-of-00189.safetensors", + "model.layers.21.mlp.up_proj.weight_scale": "model-00044-of-00189.safetensors", + "model.layers.21.post_attention_layernorm.bias": "model-00045-of-00189.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00045-of-00189.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00043-of-00189.safetensors", + "model.layers.21.self_attn.k_proj.weight_scale": "model-00043-of-00189.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00043-of-00189.safetensors", + "model.layers.21.self_attn.o_proj.weight_scale": "model-00043-of-00189.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00043-of-00189.safetensors", + "model.layers.21.self_attn.q_proj.weight_scale": "model-00043-of-00189.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00043-of-00189.safetensors", + "model.layers.21.self_attn.v_proj.weight_scale": "model-00043-of-00189.safetensors", + "model.layers.22.input_layernorm.bias": "model-00047-of-00189.safetensors", + "model.layers.22.input_layernorm.weight": "model-00047-of-00189.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00047-of-00189.safetensors", + "model.layers.22.mlp.down_proj.weight_scale": "model-00047-of-00189.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00046-of-00189.safetensors", + "model.layers.22.mlp.up_proj.weight_scale": "model-00046-of-00189.safetensors", + "model.layers.22.post_attention_layernorm.bias": "model-00047-of-00189.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00047-of-00189.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00045-of-00189.safetensors", + "model.layers.22.self_attn.k_proj.weight_scale": "model-00045-of-00189.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00045-of-00189.safetensors", + "model.layers.22.self_attn.o_proj.weight_scale": "model-00045-of-00189.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00045-of-00189.safetensors", + "model.layers.22.self_attn.q_proj.weight_scale": "model-00045-of-00189.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00045-of-00189.safetensors", + "model.layers.22.self_attn.v_proj.weight_scale": "model-00045-of-00189.safetensors", + "model.layers.23.input_layernorm.bias": "model-00049-of-00189.safetensors", + "model.layers.23.input_layernorm.weight": "model-00049-of-00189.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00049-of-00189.safetensors", + "model.layers.23.mlp.down_proj.weight_scale": "model-00049-of-00189.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00048-of-00189.safetensors", + "model.layers.23.mlp.up_proj.weight_scale": "model-00048-of-00189.safetensors", + "model.layers.23.post_attention_layernorm.bias": "model-00049-of-00189.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00049-of-00189.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00047-of-00189.safetensors", + "model.layers.23.self_attn.k_proj.weight_scale": "model-00047-of-00189.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00047-of-00189.safetensors", + "model.layers.23.self_attn.o_proj.weight_scale": "model-00047-of-00189.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00047-of-00189.safetensors", + "model.layers.23.self_attn.q_proj.weight_scale": "model-00047-of-00189.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00047-of-00189.safetensors", + "model.layers.23.self_attn.v_proj.weight_scale": "model-00047-of-00189.safetensors", + "model.layers.24.input_layernorm.bias": "model-00051-of-00189.safetensors", + "model.layers.24.input_layernorm.weight": "model-00051-of-00189.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00051-of-00189.safetensors", + "model.layers.24.mlp.down_proj.weight_scale": "model-00051-of-00189.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00050-of-00189.safetensors", + "model.layers.24.mlp.up_proj.weight_scale": "model-00050-of-00189.safetensors", + "model.layers.24.post_attention_layernorm.bias": "model-00051-of-00189.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00051-of-00189.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00049-of-00189.safetensors", + "model.layers.24.self_attn.k_proj.weight_scale": "model-00049-of-00189.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00049-of-00189.safetensors", + "model.layers.24.self_attn.o_proj.weight_scale": "model-00049-of-00189.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00049-of-00189.safetensors", + "model.layers.24.self_attn.q_proj.weight_scale": "model-00049-of-00189.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00049-of-00189.safetensors", + "model.layers.24.self_attn.v_proj.weight_scale": "model-00049-of-00189.safetensors", + "model.layers.25.input_layernorm.bias": "model-00053-of-00189.safetensors", + "model.layers.25.input_layernorm.weight": "model-00053-of-00189.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00053-of-00189.safetensors", + "model.layers.25.mlp.down_proj.weight_scale": "model-00053-of-00189.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00052-of-00189.safetensors", + "model.layers.25.mlp.up_proj.weight_scale": "model-00052-of-00189.safetensors", + "model.layers.25.post_attention_layernorm.bias": "model-00053-of-00189.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00053-of-00189.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00051-of-00189.safetensors", + "model.layers.25.self_attn.k_proj.weight_scale": "model-00051-of-00189.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00051-of-00189.safetensors", + "model.layers.25.self_attn.o_proj.weight_scale": "model-00051-of-00189.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00051-of-00189.safetensors", + "model.layers.25.self_attn.q_proj.weight_scale": "model-00051-of-00189.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00051-of-00189.safetensors", + "model.layers.25.self_attn.v_proj.weight_scale": "model-00051-of-00189.safetensors", + "model.layers.26.input_layernorm.bias": "model-00055-of-00189.safetensors", + "model.layers.26.input_layernorm.weight": "model-00055-of-00189.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00055-of-00189.safetensors", + "model.layers.26.mlp.down_proj.weight_scale": "model-00055-of-00189.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00054-of-00189.safetensors", + "model.layers.26.mlp.up_proj.weight_scale": "model-00054-of-00189.safetensors", + "model.layers.26.post_attention_layernorm.bias": "model-00055-of-00189.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00055-of-00189.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00053-of-00189.safetensors", + "model.layers.26.self_attn.k_proj.weight_scale": "model-00053-of-00189.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00053-of-00189.safetensors", + "model.layers.26.self_attn.o_proj.weight_scale": "model-00053-of-00189.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00053-of-00189.safetensors", + "model.layers.26.self_attn.q_proj.weight_scale": "model-00053-of-00189.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00053-of-00189.safetensors", + "model.layers.26.self_attn.v_proj.weight_scale": "model-00053-of-00189.safetensors", + "model.layers.27.input_layernorm.bias": "model-00057-of-00189.safetensors", + "model.layers.27.input_layernorm.weight": "model-00057-of-00189.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00057-of-00189.safetensors", + "model.layers.27.mlp.down_proj.weight_scale": "model-00057-of-00189.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00056-of-00189.safetensors", + "model.layers.27.mlp.up_proj.weight_scale": "model-00056-of-00189.safetensors", + "model.layers.27.post_attention_layernorm.bias": "model-00057-of-00189.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00057-of-00189.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00055-of-00189.safetensors", + "model.layers.27.self_attn.k_proj.weight_scale": "model-00055-of-00189.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00055-of-00189.safetensors", + "model.layers.27.self_attn.o_proj.weight_scale": "model-00055-of-00189.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00055-of-00189.safetensors", + "model.layers.27.self_attn.q_proj.weight_scale": "model-00055-of-00189.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00055-of-00189.safetensors", + "model.layers.27.self_attn.v_proj.weight_scale": "model-00055-of-00189.safetensors", + "model.layers.28.input_layernorm.bias": "model-00059-of-00189.safetensors", + "model.layers.28.input_layernorm.weight": "model-00059-of-00189.safetensors", + "model.layers.28.mlp.down_proj.weight": "model-00059-of-00189.safetensors", + "model.layers.28.mlp.down_proj.weight_scale": "model-00059-of-00189.safetensors", + "model.layers.28.mlp.up_proj.weight": "model-00058-of-00189.safetensors", + "model.layers.28.mlp.up_proj.weight_scale": "model-00058-of-00189.safetensors", + "model.layers.28.post_attention_layernorm.bias": "model-00059-of-00189.safetensors", + "model.layers.28.post_attention_layernorm.weight": "model-00059-of-00189.safetensors", + "model.layers.28.self_attn.k_proj.weight": "model-00057-of-00189.safetensors", + "model.layers.28.self_attn.k_proj.weight_scale": "model-00057-of-00189.safetensors", + "model.layers.28.self_attn.o_proj.weight": "model-00057-of-00189.safetensors", + "model.layers.28.self_attn.o_proj.weight_scale": "model-00057-of-00189.safetensors", + "model.layers.28.self_attn.q_proj.weight": "model-00057-of-00189.safetensors", + "model.layers.28.self_attn.q_proj.weight_scale": "model-00057-of-00189.safetensors", + "model.layers.28.self_attn.v_proj.weight": "model-00057-of-00189.safetensors", + "model.layers.28.self_attn.v_proj.weight_scale": "model-00057-of-00189.safetensors", + "model.layers.29.input_layernorm.bias": "model-00061-of-00189.safetensors", + "model.layers.29.input_layernorm.weight": "model-00061-of-00189.safetensors", + "model.layers.29.mlp.down_proj.weight": "model-00061-of-00189.safetensors", + "model.layers.29.mlp.down_proj.weight_scale": "model-00061-of-00189.safetensors", + "model.layers.29.mlp.up_proj.weight": "model-00060-of-00189.safetensors", + "model.layers.29.mlp.up_proj.weight_scale": "model-00060-of-00189.safetensors", + "model.layers.29.post_attention_layernorm.bias": "model-00061-of-00189.safetensors", + "model.layers.29.post_attention_layernorm.weight": "model-00061-of-00189.safetensors", + "model.layers.29.self_attn.k_proj.weight": "model-00059-of-00189.safetensors", + "model.layers.29.self_attn.k_proj.weight_scale": "model-00059-of-00189.safetensors", + "model.layers.29.self_attn.o_proj.weight": "model-00059-of-00189.safetensors", + "model.layers.29.self_attn.o_proj.weight_scale": "model-00059-of-00189.safetensors", + "model.layers.29.self_attn.q_proj.weight": "model-00059-of-00189.safetensors", + "model.layers.29.self_attn.q_proj.weight_scale": "model-00059-of-00189.safetensors", + "model.layers.29.self_attn.v_proj.weight": "model-00059-of-00189.safetensors", + "model.layers.29.self_attn.v_proj.weight_scale": "model-00059-of-00189.safetensors", + "model.layers.3.input_layernorm.bias": "model-00009-of-00189.safetensors", + "model.layers.3.input_layernorm.weight": "model-00009-of-00189.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00009-of-00189.safetensors", + "model.layers.3.mlp.down_proj.weight_scale": "model-00009-of-00189.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00008-of-00189.safetensors", + "model.layers.3.mlp.up_proj.weight_scale": "model-00008-of-00189.safetensors", + "model.layers.3.post_attention_layernorm.bias": "model-00009-of-00189.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00009-of-00189.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00007-of-00189.safetensors", + "model.layers.3.self_attn.k_proj.weight_scale": "model-00007-of-00189.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00007-of-00189.safetensors", + "model.layers.3.self_attn.o_proj.weight_scale": "model-00007-of-00189.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00007-of-00189.safetensors", + "model.layers.3.self_attn.q_proj.weight_scale": "model-00007-of-00189.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00007-of-00189.safetensors", + "model.layers.3.self_attn.v_proj.weight_scale": "model-00007-of-00189.safetensors", + "model.layers.30.input_layernorm.bias": "model-00063-of-00189.safetensors", + "model.layers.30.input_layernorm.weight": "model-00063-of-00189.safetensors", + "model.layers.30.mlp.down_proj.weight": "model-00063-of-00189.safetensors", + "model.layers.30.mlp.down_proj.weight_scale": "model-00063-of-00189.safetensors", + "model.layers.30.mlp.up_proj.weight": "model-00062-of-00189.safetensors", + "model.layers.30.mlp.up_proj.weight_scale": "model-00062-of-00189.safetensors", + "model.layers.30.post_attention_layernorm.bias": "model-00063-of-00189.safetensors", + "model.layers.30.post_attention_layernorm.weight": "model-00063-of-00189.safetensors", + "model.layers.30.self_attn.k_proj.weight": "model-00061-of-00189.safetensors", + "model.layers.30.self_attn.k_proj.weight_scale": "model-00061-of-00189.safetensors", + "model.layers.30.self_attn.o_proj.weight": "model-00061-of-00189.safetensors", + "model.layers.30.self_attn.o_proj.weight_scale": "model-00061-of-00189.safetensors", + "model.layers.30.self_attn.q_proj.weight": "model-00061-of-00189.safetensors", + "model.layers.30.self_attn.q_proj.weight_scale": "model-00061-of-00189.safetensors", + "model.layers.30.self_attn.v_proj.weight": "model-00061-of-00189.safetensors", + "model.layers.30.self_attn.v_proj.weight_scale": "model-00061-of-00189.safetensors", + "model.layers.31.input_layernorm.bias": "model-00065-of-00189.safetensors", + "model.layers.31.input_layernorm.weight": "model-00065-of-00189.safetensors", + "model.layers.31.mlp.down_proj.weight": "model-00065-of-00189.safetensors", + "model.layers.31.mlp.down_proj.weight_scale": "model-00065-of-00189.safetensors", + "model.layers.31.mlp.up_proj.weight": "model-00064-of-00189.safetensors", + "model.layers.31.mlp.up_proj.weight_scale": "model-00064-of-00189.safetensors", + "model.layers.31.post_attention_layernorm.bias": "model-00065-of-00189.safetensors", + "model.layers.31.post_attention_layernorm.weight": "model-00065-of-00189.safetensors", + "model.layers.31.self_attn.k_proj.weight": "model-00063-of-00189.safetensors", + "model.layers.31.self_attn.k_proj.weight_scale": "model-00063-of-00189.safetensors", + "model.layers.31.self_attn.o_proj.weight": "model-00063-of-00189.safetensors", + "model.layers.31.self_attn.o_proj.weight_scale": "model-00063-of-00189.safetensors", + "model.layers.31.self_attn.q_proj.weight": "model-00063-of-00189.safetensors", + "model.layers.31.self_attn.q_proj.weight_scale": "model-00063-of-00189.safetensors", + "model.layers.31.self_attn.v_proj.weight": "model-00063-of-00189.safetensors", + "model.layers.31.self_attn.v_proj.weight_scale": "model-00063-of-00189.safetensors", + "model.layers.32.input_layernorm.bias": "model-00067-of-00189.safetensors", + "model.layers.32.input_layernorm.weight": "model-00067-of-00189.safetensors", + "model.layers.32.mlp.down_proj.weight": "model-00067-of-00189.safetensors", + "model.layers.32.mlp.down_proj.weight_scale": "model-00067-of-00189.safetensors", + "model.layers.32.mlp.up_proj.weight": "model-00066-of-00189.safetensors", + "model.layers.32.mlp.up_proj.weight_scale": "model-00066-of-00189.safetensors", + "model.layers.32.post_attention_layernorm.bias": "model-00067-of-00189.safetensors", + "model.layers.32.post_attention_layernorm.weight": "model-00067-of-00189.safetensors", + "model.layers.32.self_attn.k_proj.weight": "model-00065-of-00189.safetensors", + "model.layers.32.self_attn.k_proj.weight_scale": "model-00065-of-00189.safetensors", + "model.layers.32.self_attn.o_proj.weight": "model-00065-of-00189.safetensors", + "model.layers.32.self_attn.o_proj.weight_scale": "model-00065-of-00189.safetensors", + "model.layers.32.self_attn.q_proj.weight": "model-00065-of-00189.safetensors", + "model.layers.32.self_attn.q_proj.weight_scale": "model-00065-of-00189.safetensors", + "model.layers.32.self_attn.v_proj.weight": "model-00065-of-00189.safetensors", + "model.layers.32.self_attn.v_proj.weight_scale": "model-00065-of-00189.safetensors", + "model.layers.33.input_layernorm.bias": "model-00069-of-00189.safetensors", + "model.layers.33.input_layernorm.weight": "model-00069-of-00189.safetensors", + "model.layers.33.mlp.down_proj.weight": "model-00069-of-00189.safetensors", + "model.layers.33.mlp.down_proj.weight_scale": "model-00069-of-00189.safetensors", + "model.layers.33.mlp.up_proj.weight": "model-00068-of-00189.safetensors", + "model.layers.33.mlp.up_proj.weight_scale": "model-00068-of-00189.safetensors", + "model.layers.33.post_attention_layernorm.bias": "model-00069-of-00189.safetensors", + "model.layers.33.post_attention_layernorm.weight": "model-00069-of-00189.safetensors", + "model.layers.33.self_attn.k_proj.weight": "model-00067-of-00189.safetensors", + "model.layers.33.self_attn.k_proj.weight_scale": "model-00067-of-00189.safetensors", + "model.layers.33.self_attn.o_proj.weight": "model-00067-of-00189.safetensors", + "model.layers.33.self_attn.o_proj.weight_scale": "model-00067-of-00189.safetensors", + "model.layers.33.self_attn.q_proj.weight": "model-00067-of-00189.safetensors", + "model.layers.33.self_attn.q_proj.weight_scale": "model-00067-of-00189.safetensors", + "model.layers.33.self_attn.v_proj.weight": "model-00067-of-00189.safetensors", + "model.layers.33.self_attn.v_proj.weight_scale": "model-00067-of-00189.safetensors", + "model.layers.34.input_layernorm.bias": "model-00071-of-00189.safetensors", + "model.layers.34.input_layernorm.weight": "model-00071-of-00189.safetensors", + "model.layers.34.mlp.down_proj.weight": "model-00071-of-00189.safetensors", + "model.layers.34.mlp.down_proj.weight_scale": "model-00071-of-00189.safetensors", + "model.layers.34.mlp.up_proj.weight": "model-00070-of-00189.safetensors", + "model.layers.34.mlp.up_proj.weight_scale": "model-00070-of-00189.safetensors", + "model.layers.34.post_attention_layernorm.bias": "model-00071-of-00189.safetensors", + "model.layers.34.post_attention_layernorm.weight": "model-00071-of-00189.safetensors", + "model.layers.34.self_attn.k_proj.weight": "model-00069-of-00189.safetensors", + "model.layers.34.self_attn.k_proj.weight_scale": "model-00069-of-00189.safetensors", + "model.layers.34.self_attn.o_proj.weight": "model-00069-of-00189.safetensors", + "model.layers.34.self_attn.o_proj.weight_scale": "model-00069-of-00189.safetensors", + "model.layers.34.self_attn.q_proj.weight": "model-00069-of-00189.safetensors", + "model.layers.34.self_attn.q_proj.weight_scale": "model-00069-of-00189.safetensors", + "model.layers.34.self_attn.v_proj.weight": "model-00069-of-00189.safetensors", + "model.layers.34.self_attn.v_proj.weight_scale": "model-00069-of-00189.safetensors", + "model.layers.35.input_layernorm.bias": "model-00073-of-00189.safetensors", + "model.layers.35.input_layernorm.weight": "model-00073-of-00189.safetensors", + "model.layers.35.mlp.down_proj.weight": "model-00073-of-00189.safetensors", + "model.layers.35.mlp.down_proj.weight_scale": "model-00073-of-00189.safetensors", + "model.layers.35.mlp.up_proj.weight": "model-00072-of-00189.safetensors", + "model.layers.35.mlp.up_proj.weight_scale": "model-00072-of-00189.safetensors", + "model.layers.35.post_attention_layernorm.bias": "model-00073-of-00189.safetensors", + "model.layers.35.post_attention_layernorm.weight": "model-00073-of-00189.safetensors", + "model.layers.35.self_attn.k_proj.weight": "model-00071-of-00189.safetensors", + "model.layers.35.self_attn.k_proj.weight_scale": "model-00071-of-00189.safetensors", + "model.layers.35.self_attn.o_proj.weight": "model-00071-of-00189.safetensors", + "model.layers.35.self_attn.o_proj.weight_scale": "model-00071-of-00189.safetensors", + "model.layers.35.self_attn.q_proj.weight": "model-00071-of-00189.safetensors", + "model.layers.35.self_attn.q_proj.weight_scale": "model-00071-of-00189.safetensors", + "model.layers.35.self_attn.v_proj.weight": "model-00071-of-00189.safetensors", + "model.layers.35.self_attn.v_proj.weight_scale": "model-00071-of-00189.safetensors", + "model.layers.36.input_layernorm.bias": "model-00075-of-00189.safetensors", + "model.layers.36.input_layernorm.weight": "model-00075-of-00189.safetensors", + "model.layers.36.mlp.down_proj.weight": "model-00075-of-00189.safetensors", + "model.layers.36.mlp.down_proj.weight_scale": "model-00075-of-00189.safetensors", + "model.layers.36.mlp.up_proj.weight": "model-00074-of-00189.safetensors", + "model.layers.36.mlp.up_proj.weight_scale": "model-00074-of-00189.safetensors", + "model.layers.36.post_attention_layernorm.bias": "model-00075-of-00189.safetensors", + "model.layers.36.post_attention_layernorm.weight": "model-00075-of-00189.safetensors", + "model.layers.36.self_attn.k_proj.weight": "model-00073-of-00189.safetensors", + "model.layers.36.self_attn.k_proj.weight_scale": "model-00073-of-00189.safetensors", + "model.layers.36.self_attn.o_proj.weight": "model-00073-of-00189.safetensors", + "model.layers.36.self_attn.o_proj.weight_scale": "model-00073-of-00189.safetensors", + "model.layers.36.self_attn.q_proj.weight": "model-00073-of-00189.safetensors", + "model.layers.36.self_attn.q_proj.weight_scale": "model-00073-of-00189.safetensors", + "model.layers.36.self_attn.v_proj.weight": "model-00073-of-00189.safetensors", + "model.layers.36.self_attn.v_proj.weight_scale": "model-00073-of-00189.safetensors", + "model.layers.37.input_layernorm.bias": "model-00077-of-00189.safetensors", + "model.layers.37.input_layernorm.weight": "model-00077-of-00189.safetensors", + "model.layers.37.mlp.down_proj.weight": "model-00077-of-00189.safetensors", + "model.layers.37.mlp.down_proj.weight_scale": "model-00077-of-00189.safetensors", + "model.layers.37.mlp.up_proj.weight": "model-00076-of-00189.safetensors", + "model.layers.37.mlp.up_proj.weight_scale": "model-00076-of-00189.safetensors", + "model.layers.37.post_attention_layernorm.bias": "model-00077-of-00189.safetensors", + "model.layers.37.post_attention_layernorm.weight": "model-00077-of-00189.safetensors", + "model.layers.37.self_attn.k_proj.weight": "model-00075-of-00189.safetensors", + "model.layers.37.self_attn.k_proj.weight_scale": "model-00075-of-00189.safetensors", + "model.layers.37.self_attn.o_proj.weight": "model-00075-of-00189.safetensors", + "model.layers.37.self_attn.o_proj.weight_scale": "model-00075-of-00189.safetensors", + "model.layers.37.self_attn.q_proj.weight": "model-00075-of-00189.safetensors", + "model.layers.37.self_attn.q_proj.weight_scale": "model-00075-of-00189.safetensors", + "model.layers.37.self_attn.v_proj.weight": "model-00075-of-00189.safetensors", + "model.layers.37.self_attn.v_proj.weight_scale": "model-00075-of-00189.safetensors", + "model.layers.38.input_layernorm.bias": "model-00079-of-00189.safetensors", + "model.layers.38.input_layernorm.weight": "model-00079-of-00189.safetensors", + "model.layers.38.mlp.down_proj.weight": "model-00079-of-00189.safetensors", + "model.layers.38.mlp.down_proj.weight_scale": "model-00079-of-00189.safetensors", + "model.layers.38.mlp.up_proj.weight": "model-00078-of-00189.safetensors", + "model.layers.38.mlp.up_proj.weight_scale": "model-00078-of-00189.safetensors", + "model.layers.38.post_attention_layernorm.bias": "model-00079-of-00189.safetensors", + "model.layers.38.post_attention_layernorm.weight": "model-00079-of-00189.safetensors", + "model.layers.38.self_attn.k_proj.weight": "model-00077-of-00189.safetensors", + "model.layers.38.self_attn.k_proj.weight_scale": "model-00077-of-00189.safetensors", + "model.layers.38.self_attn.o_proj.weight": "model-00077-of-00189.safetensors", + "model.layers.38.self_attn.o_proj.weight_scale": "model-00077-of-00189.safetensors", + "model.layers.38.self_attn.q_proj.weight": "model-00077-of-00189.safetensors", + "model.layers.38.self_attn.q_proj.weight_scale": "model-00077-of-00189.safetensors", + "model.layers.38.self_attn.v_proj.weight": "model-00077-of-00189.safetensors", + "model.layers.38.self_attn.v_proj.weight_scale": "model-00077-of-00189.safetensors", + "model.layers.39.input_layernorm.bias": "model-00081-of-00189.safetensors", + "model.layers.39.input_layernorm.weight": "model-00081-of-00189.safetensors", + "model.layers.39.mlp.down_proj.weight": "model-00081-of-00189.safetensors", + "model.layers.39.mlp.down_proj.weight_scale": "model-00081-of-00189.safetensors", + "model.layers.39.mlp.up_proj.weight": "model-00080-of-00189.safetensors", + "model.layers.39.mlp.up_proj.weight_scale": "model-00080-of-00189.safetensors", + "model.layers.39.post_attention_layernorm.bias": "model-00081-of-00189.safetensors", + "model.layers.39.post_attention_layernorm.weight": "model-00081-of-00189.safetensors", + "model.layers.39.self_attn.k_proj.weight": "model-00079-of-00189.safetensors", + "model.layers.39.self_attn.k_proj.weight_scale": "model-00079-of-00189.safetensors", + "model.layers.39.self_attn.o_proj.weight": "model-00079-of-00189.safetensors", + "model.layers.39.self_attn.o_proj.weight_scale": "model-00079-of-00189.safetensors", + "model.layers.39.self_attn.q_proj.weight": "model-00079-of-00189.safetensors", + "model.layers.39.self_attn.q_proj.weight_scale": "model-00079-of-00189.safetensors", + "model.layers.39.self_attn.v_proj.weight": "model-00079-of-00189.safetensors", + "model.layers.39.self_attn.v_proj.weight_scale": "model-00079-of-00189.safetensors", + "model.layers.4.input_layernorm.bias": "model-00011-of-00189.safetensors", + "model.layers.4.input_layernorm.weight": "model-00011-of-00189.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00011-of-00189.safetensors", + "model.layers.4.mlp.down_proj.weight_scale": "model-00011-of-00189.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00010-of-00189.safetensors", + "model.layers.4.mlp.up_proj.weight_scale": "model-00010-of-00189.safetensors", + "model.layers.4.post_attention_layernorm.bias": "model-00011-of-00189.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00011-of-00189.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00009-of-00189.safetensors", + "model.layers.4.self_attn.k_proj.weight_scale": "model-00009-of-00189.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00009-of-00189.safetensors", + "model.layers.4.self_attn.o_proj.weight_scale": "model-00009-of-00189.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00009-of-00189.safetensors", + "model.layers.4.self_attn.q_proj.weight_scale": "model-00009-of-00189.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00009-of-00189.safetensors", + "model.layers.4.self_attn.v_proj.weight_scale": "model-00009-of-00189.safetensors", + "model.layers.40.input_layernorm.bias": "model-00083-of-00189.safetensors", + "model.layers.40.input_layernorm.weight": "model-00083-of-00189.safetensors", + "model.layers.40.mlp.down_proj.weight": "model-00083-of-00189.safetensors", + "model.layers.40.mlp.down_proj.weight_scale": "model-00083-of-00189.safetensors", + "model.layers.40.mlp.up_proj.weight": "model-00082-of-00189.safetensors", + "model.layers.40.mlp.up_proj.weight_scale": "model-00082-of-00189.safetensors", + "model.layers.40.post_attention_layernorm.bias": "model-00083-of-00189.safetensors", + "model.layers.40.post_attention_layernorm.weight": "model-00083-of-00189.safetensors", + "model.layers.40.self_attn.k_proj.weight": "model-00081-of-00189.safetensors", + "model.layers.40.self_attn.k_proj.weight_scale": "model-00081-of-00189.safetensors", + "model.layers.40.self_attn.o_proj.weight": "model-00081-of-00189.safetensors", + "model.layers.40.self_attn.o_proj.weight_scale": "model-00081-of-00189.safetensors", + "model.layers.40.self_attn.q_proj.weight": "model-00081-of-00189.safetensors", + "model.layers.40.self_attn.q_proj.weight_scale": "model-00081-of-00189.safetensors", + "model.layers.40.self_attn.v_proj.weight": "model-00081-of-00189.safetensors", + "model.layers.40.self_attn.v_proj.weight_scale": "model-00081-of-00189.safetensors", + "model.layers.41.input_layernorm.bias": "model-00085-of-00189.safetensors", + "model.layers.41.input_layernorm.weight": "model-00085-of-00189.safetensors", + "model.layers.41.mlp.down_proj.weight": "model-00085-of-00189.safetensors", + "model.layers.41.mlp.down_proj.weight_scale": "model-00085-of-00189.safetensors", + "model.layers.41.mlp.up_proj.weight": "model-00084-of-00189.safetensors", + "model.layers.41.mlp.up_proj.weight_scale": "model-00084-of-00189.safetensors", + "model.layers.41.post_attention_layernorm.bias": "model-00085-of-00189.safetensors", + "model.layers.41.post_attention_layernorm.weight": "model-00085-of-00189.safetensors", + "model.layers.41.self_attn.k_proj.weight": "model-00083-of-00189.safetensors", + "model.layers.41.self_attn.k_proj.weight_scale": "model-00083-of-00189.safetensors", + "model.layers.41.self_attn.o_proj.weight": "model-00083-of-00189.safetensors", + "model.layers.41.self_attn.o_proj.weight_scale": "model-00083-of-00189.safetensors", + "model.layers.41.self_attn.q_proj.weight": "model-00083-of-00189.safetensors", + "model.layers.41.self_attn.q_proj.weight_scale": "model-00083-of-00189.safetensors", + "model.layers.41.self_attn.v_proj.weight": "model-00083-of-00189.safetensors", + "model.layers.41.self_attn.v_proj.weight_scale": "model-00083-of-00189.safetensors", + "model.layers.42.input_layernorm.bias": "model-00087-of-00189.safetensors", + "model.layers.42.input_layernorm.weight": "model-00087-of-00189.safetensors", + "model.layers.42.mlp.down_proj.weight": "model-00087-of-00189.safetensors", + "model.layers.42.mlp.down_proj.weight_scale": "model-00087-of-00189.safetensors", + "model.layers.42.mlp.up_proj.weight": "model-00086-of-00189.safetensors", + "model.layers.42.mlp.up_proj.weight_scale": "model-00086-of-00189.safetensors", + "model.layers.42.post_attention_layernorm.bias": "model-00087-of-00189.safetensors", + "model.layers.42.post_attention_layernorm.weight": "model-00087-of-00189.safetensors", + "model.layers.42.self_attn.k_proj.weight": "model-00085-of-00189.safetensors", + "model.layers.42.self_attn.k_proj.weight_scale": "model-00085-of-00189.safetensors", + "model.layers.42.self_attn.o_proj.weight": "model-00085-of-00189.safetensors", + "model.layers.42.self_attn.o_proj.weight_scale": "model-00085-of-00189.safetensors", + "model.layers.42.self_attn.q_proj.weight": "model-00085-of-00189.safetensors", + "model.layers.42.self_attn.q_proj.weight_scale": "model-00085-of-00189.safetensors", + "model.layers.42.self_attn.v_proj.weight": "model-00085-of-00189.safetensors", + "model.layers.42.self_attn.v_proj.weight_scale": "model-00085-of-00189.safetensors", + "model.layers.43.input_layernorm.bias": "model-00089-of-00189.safetensors", + "model.layers.43.input_layernorm.weight": "model-00089-of-00189.safetensors", + "model.layers.43.mlp.down_proj.weight": "model-00089-of-00189.safetensors", + "model.layers.43.mlp.down_proj.weight_scale": "model-00089-of-00189.safetensors", + "model.layers.43.mlp.up_proj.weight": "model-00088-of-00189.safetensors", + "model.layers.43.mlp.up_proj.weight_scale": "model-00088-of-00189.safetensors", + "model.layers.43.post_attention_layernorm.bias": "model-00089-of-00189.safetensors", + "model.layers.43.post_attention_layernorm.weight": "model-00089-of-00189.safetensors", + "model.layers.43.self_attn.k_proj.weight": "model-00087-of-00189.safetensors", + "model.layers.43.self_attn.k_proj.weight_scale": "model-00087-of-00189.safetensors", + "model.layers.43.self_attn.o_proj.weight": "model-00087-of-00189.safetensors", + "model.layers.43.self_attn.o_proj.weight_scale": "model-00087-of-00189.safetensors", + "model.layers.43.self_attn.q_proj.weight": "model-00087-of-00189.safetensors", + "model.layers.43.self_attn.q_proj.weight_scale": "model-00087-of-00189.safetensors", + "model.layers.43.self_attn.v_proj.weight": "model-00087-of-00189.safetensors", + "model.layers.43.self_attn.v_proj.weight_scale": "model-00087-of-00189.safetensors", + "model.layers.44.input_layernorm.bias": "model-00091-of-00189.safetensors", + "model.layers.44.input_layernorm.weight": "model-00091-of-00189.safetensors", + "model.layers.44.mlp.down_proj.weight": "model-00091-of-00189.safetensors", + "model.layers.44.mlp.down_proj.weight_scale": "model-00091-of-00189.safetensors", + "model.layers.44.mlp.up_proj.weight": "model-00090-of-00189.safetensors", + "model.layers.44.mlp.up_proj.weight_scale": "model-00090-of-00189.safetensors", + "model.layers.44.post_attention_layernorm.bias": "model-00091-of-00189.safetensors", + "model.layers.44.post_attention_layernorm.weight": "model-00091-of-00189.safetensors", + "model.layers.44.self_attn.k_proj.weight": "model-00089-of-00189.safetensors", + "model.layers.44.self_attn.k_proj.weight_scale": "model-00089-of-00189.safetensors", + "model.layers.44.self_attn.o_proj.weight": "model-00089-of-00189.safetensors", + "model.layers.44.self_attn.o_proj.weight_scale": "model-00089-of-00189.safetensors", + "model.layers.44.self_attn.q_proj.weight": "model-00089-of-00189.safetensors", + "model.layers.44.self_attn.q_proj.weight_scale": "model-00089-of-00189.safetensors", + "model.layers.44.self_attn.v_proj.weight": "model-00089-of-00189.safetensors", + "model.layers.44.self_attn.v_proj.weight_scale": "model-00089-of-00189.safetensors", + "model.layers.45.input_layernorm.bias": "model-00093-of-00189.safetensors", + "model.layers.45.input_layernorm.weight": "model-00093-of-00189.safetensors", + "model.layers.45.mlp.down_proj.weight": "model-00093-of-00189.safetensors", + "model.layers.45.mlp.down_proj.weight_scale": "model-00093-of-00189.safetensors", + "model.layers.45.mlp.up_proj.weight": "model-00092-of-00189.safetensors", + "model.layers.45.mlp.up_proj.weight_scale": "model-00092-of-00189.safetensors", + "model.layers.45.post_attention_layernorm.bias": "model-00093-of-00189.safetensors", + "model.layers.45.post_attention_layernorm.weight": "model-00093-of-00189.safetensors", + "model.layers.45.self_attn.k_proj.weight": "model-00091-of-00189.safetensors", + "model.layers.45.self_attn.k_proj.weight_scale": "model-00091-of-00189.safetensors", + "model.layers.45.self_attn.o_proj.weight": "model-00091-of-00189.safetensors", + "model.layers.45.self_attn.o_proj.weight_scale": "model-00091-of-00189.safetensors", + "model.layers.45.self_attn.q_proj.weight": "model-00091-of-00189.safetensors", + "model.layers.45.self_attn.q_proj.weight_scale": "model-00091-of-00189.safetensors", + "model.layers.45.self_attn.v_proj.weight": "model-00091-of-00189.safetensors", + "model.layers.45.self_attn.v_proj.weight_scale": "model-00091-of-00189.safetensors", + "model.layers.46.input_layernorm.bias": "model-00095-of-00189.safetensors", + "model.layers.46.input_layernorm.weight": "model-00095-of-00189.safetensors", + "model.layers.46.mlp.down_proj.weight": "model-00095-of-00189.safetensors", + "model.layers.46.mlp.down_proj.weight_scale": "model-00095-of-00189.safetensors", + "model.layers.46.mlp.up_proj.weight": "model-00094-of-00189.safetensors", + "model.layers.46.mlp.up_proj.weight_scale": "model-00094-of-00189.safetensors", + "model.layers.46.post_attention_layernorm.bias": "model-00095-of-00189.safetensors", + "model.layers.46.post_attention_layernorm.weight": "model-00095-of-00189.safetensors", + "model.layers.46.self_attn.k_proj.weight": "model-00093-of-00189.safetensors", + "model.layers.46.self_attn.k_proj.weight_scale": "model-00093-of-00189.safetensors", + "model.layers.46.self_attn.o_proj.weight": "model-00093-of-00189.safetensors", + "model.layers.46.self_attn.o_proj.weight_scale": "model-00093-of-00189.safetensors", + "model.layers.46.self_attn.q_proj.weight": "model-00093-of-00189.safetensors", + "model.layers.46.self_attn.q_proj.weight_scale": "model-00093-of-00189.safetensors", + "model.layers.46.self_attn.v_proj.weight": "model-00093-of-00189.safetensors", + "model.layers.46.self_attn.v_proj.weight_scale": "model-00093-of-00189.safetensors", + "model.layers.47.input_layernorm.bias": "model-00097-of-00189.safetensors", + "model.layers.47.input_layernorm.weight": "model-00097-of-00189.safetensors", + "model.layers.47.mlp.down_proj.weight": "model-00097-of-00189.safetensors", + "model.layers.47.mlp.down_proj.weight_scale": "model-00097-of-00189.safetensors", + "model.layers.47.mlp.up_proj.weight": "model-00096-of-00189.safetensors", + "model.layers.47.mlp.up_proj.weight_scale": "model-00096-of-00189.safetensors", + "model.layers.47.post_attention_layernorm.bias": "model-00097-of-00189.safetensors", + "model.layers.47.post_attention_layernorm.weight": "model-00097-of-00189.safetensors", + "model.layers.47.self_attn.k_proj.weight": "model-00095-of-00189.safetensors", + "model.layers.47.self_attn.k_proj.weight_scale": "model-00095-of-00189.safetensors", + "model.layers.47.self_attn.o_proj.weight": "model-00095-of-00189.safetensors", + "model.layers.47.self_attn.o_proj.weight_scale": "model-00095-of-00189.safetensors", + "model.layers.47.self_attn.q_proj.weight": "model-00095-of-00189.safetensors", + "model.layers.47.self_attn.q_proj.weight_scale": "model-00095-of-00189.safetensors", + "model.layers.47.self_attn.v_proj.weight": "model-00095-of-00189.safetensors", + "model.layers.47.self_attn.v_proj.weight_scale": "model-00095-of-00189.safetensors", + "model.layers.48.input_layernorm.bias": "model-00099-of-00189.safetensors", + "model.layers.48.input_layernorm.weight": "model-00099-of-00189.safetensors", + "model.layers.48.mlp.down_proj.weight": "model-00099-of-00189.safetensors", + "model.layers.48.mlp.down_proj.weight_scale": "model-00099-of-00189.safetensors", + "model.layers.48.mlp.up_proj.weight": "model-00098-of-00189.safetensors", + "model.layers.48.mlp.up_proj.weight_scale": "model-00098-of-00189.safetensors", + "model.layers.48.post_attention_layernorm.bias": "model-00099-of-00189.safetensors", + "model.layers.48.post_attention_layernorm.weight": "model-00099-of-00189.safetensors", + "model.layers.48.self_attn.k_proj.weight": "model-00097-of-00189.safetensors", + "model.layers.48.self_attn.k_proj.weight_scale": "model-00097-of-00189.safetensors", + "model.layers.48.self_attn.o_proj.weight": "model-00097-of-00189.safetensors", + "model.layers.48.self_attn.o_proj.weight_scale": "model-00097-of-00189.safetensors", + "model.layers.48.self_attn.q_proj.weight": "model-00097-of-00189.safetensors", + "model.layers.48.self_attn.q_proj.weight_scale": "model-00097-of-00189.safetensors", + "model.layers.48.self_attn.v_proj.weight": "model-00097-of-00189.safetensors", + "model.layers.48.self_attn.v_proj.weight_scale": "model-00097-of-00189.safetensors", + "model.layers.49.input_layernorm.bias": "model-00101-of-00189.safetensors", + "model.layers.49.input_layernorm.weight": "model-00101-of-00189.safetensors", + "model.layers.49.mlp.down_proj.weight": "model-00101-of-00189.safetensors", + "model.layers.49.mlp.down_proj.weight_scale": "model-00101-of-00189.safetensors", + "model.layers.49.mlp.up_proj.weight": "model-00100-of-00189.safetensors", + "model.layers.49.mlp.up_proj.weight_scale": "model-00100-of-00189.safetensors", + "model.layers.49.post_attention_layernorm.bias": "model-00101-of-00189.safetensors", + "model.layers.49.post_attention_layernorm.weight": "model-00101-of-00189.safetensors", + "model.layers.49.self_attn.k_proj.weight": "model-00099-of-00189.safetensors", + "model.layers.49.self_attn.k_proj.weight_scale": "model-00099-of-00189.safetensors", + "model.layers.49.self_attn.o_proj.weight": "model-00099-of-00189.safetensors", + "model.layers.49.self_attn.o_proj.weight_scale": "model-00099-of-00189.safetensors", + "model.layers.49.self_attn.q_proj.weight": "model-00099-of-00189.safetensors", + "model.layers.49.self_attn.q_proj.weight_scale": "model-00099-of-00189.safetensors", + "model.layers.49.self_attn.v_proj.weight": "model-00099-of-00189.safetensors", + "model.layers.49.self_attn.v_proj.weight_scale": "model-00099-of-00189.safetensors", + "model.layers.5.input_layernorm.bias": "model-00013-of-00189.safetensors", + "model.layers.5.input_layernorm.weight": "model-00013-of-00189.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00013-of-00189.safetensors", + "model.layers.5.mlp.down_proj.weight_scale": "model-00013-of-00189.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00012-of-00189.safetensors", + "model.layers.5.mlp.up_proj.weight_scale": "model-00012-of-00189.safetensors", + "model.layers.5.post_attention_layernorm.bias": "model-00013-of-00189.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00013-of-00189.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00011-of-00189.safetensors", + "model.layers.5.self_attn.k_proj.weight_scale": "model-00011-of-00189.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00011-of-00189.safetensors", + "model.layers.5.self_attn.o_proj.weight_scale": "model-00011-of-00189.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00011-of-00189.safetensors", + "model.layers.5.self_attn.q_proj.weight_scale": "model-00011-of-00189.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00011-of-00189.safetensors", + "model.layers.5.self_attn.v_proj.weight_scale": "model-00011-of-00189.safetensors", + "model.layers.50.input_layernorm.bias": "model-00103-of-00189.safetensors", + "model.layers.50.input_layernorm.weight": "model-00103-of-00189.safetensors", + "model.layers.50.mlp.down_proj.weight": "model-00103-of-00189.safetensors", + "model.layers.50.mlp.down_proj.weight_scale": "model-00103-of-00189.safetensors", + "model.layers.50.mlp.up_proj.weight": "model-00102-of-00189.safetensors", + "model.layers.50.mlp.up_proj.weight_scale": "model-00102-of-00189.safetensors", + "model.layers.50.post_attention_layernorm.bias": "model-00103-of-00189.safetensors", + "model.layers.50.post_attention_layernorm.weight": "model-00103-of-00189.safetensors", + "model.layers.50.self_attn.k_proj.weight": "model-00101-of-00189.safetensors", + "model.layers.50.self_attn.k_proj.weight_scale": "model-00101-of-00189.safetensors", + "model.layers.50.self_attn.o_proj.weight": "model-00101-of-00189.safetensors", + "model.layers.50.self_attn.o_proj.weight_scale": "model-00101-of-00189.safetensors", + "model.layers.50.self_attn.q_proj.weight": "model-00101-of-00189.safetensors", + "model.layers.50.self_attn.q_proj.weight_scale": "model-00101-of-00189.safetensors", + "model.layers.50.self_attn.v_proj.weight": "model-00101-of-00189.safetensors", + "model.layers.50.self_attn.v_proj.weight_scale": "model-00101-of-00189.safetensors", + "model.layers.51.input_layernorm.bias": "model-00105-of-00189.safetensors", + "model.layers.51.input_layernorm.weight": "model-00105-of-00189.safetensors", + "model.layers.51.mlp.down_proj.weight": "model-00105-of-00189.safetensors", + "model.layers.51.mlp.down_proj.weight_scale": "model-00105-of-00189.safetensors", + "model.layers.51.mlp.up_proj.weight": "model-00104-of-00189.safetensors", + "model.layers.51.mlp.up_proj.weight_scale": "model-00104-of-00189.safetensors", + "model.layers.51.post_attention_layernorm.bias": "model-00105-of-00189.safetensors", + "model.layers.51.post_attention_layernorm.weight": "model-00105-of-00189.safetensors", + "model.layers.51.self_attn.k_proj.weight": "model-00103-of-00189.safetensors", + "model.layers.51.self_attn.k_proj.weight_scale": "model-00103-of-00189.safetensors", + "model.layers.51.self_attn.o_proj.weight": "model-00103-of-00189.safetensors", + "model.layers.51.self_attn.o_proj.weight_scale": "model-00103-of-00189.safetensors", + "model.layers.51.self_attn.q_proj.weight": "model-00103-of-00189.safetensors", + "model.layers.51.self_attn.q_proj.weight_scale": "model-00103-of-00189.safetensors", + "model.layers.51.self_attn.v_proj.weight": "model-00103-of-00189.safetensors", + "model.layers.51.self_attn.v_proj.weight_scale": "model-00103-of-00189.safetensors", + "model.layers.52.input_layernorm.bias": "model-00107-of-00189.safetensors", + "model.layers.52.input_layernorm.weight": "model-00107-of-00189.safetensors", + "model.layers.52.mlp.down_proj.weight": "model-00107-of-00189.safetensors", + "model.layers.52.mlp.down_proj.weight_scale": "model-00107-of-00189.safetensors", + "model.layers.52.mlp.up_proj.weight": "model-00106-of-00189.safetensors", + "model.layers.52.mlp.up_proj.weight_scale": "model-00106-of-00189.safetensors", + "model.layers.52.post_attention_layernorm.bias": "model-00107-of-00189.safetensors", + "model.layers.52.post_attention_layernorm.weight": "model-00107-of-00189.safetensors", + "model.layers.52.self_attn.k_proj.weight": "model-00105-of-00189.safetensors", + "model.layers.52.self_attn.k_proj.weight_scale": "model-00105-of-00189.safetensors", + "model.layers.52.self_attn.o_proj.weight": "model-00105-of-00189.safetensors", + "model.layers.52.self_attn.o_proj.weight_scale": "model-00105-of-00189.safetensors", + "model.layers.52.self_attn.q_proj.weight": "model-00105-of-00189.safetensors", + "model.layers.52.self_attn.q_proj.weight_scale": "model-00105-of-00189.safetensors", + "model.layers.52.self_attn.v_proj.weight": "model-00105-of-00189.safetensors", + "model.layers.52.self_attn.v_proj.weight_scale": "model-00105-of-00189.safetensors", + "model.layers.53.input_layernorm.bias": "model-00109-of-00189.safetensors", + "model.layers.53.input_layernorm.weight": "model-00109-of-00189.safetensors", + "model.layers.53.mlp.down_proj.weight": "model-00109-of-00189.safetensors", + "model.layers.53.mlp.down_proj.weight_scale": "model-00109-of-00189.safetensors", + "model.layers.53.mlp.up_proj.weight": "model-00108-of-00189.safetensors", + "model.layers.53.mlp.up_proj.weight_scale": "model-00108-of-00189.safetensors", + "model.layers.53.post_attention_layernorm.bias": "model-00109-of-00189.safetensors", + "model.layers.53.post_attention_layernorm.weight": "model-00109-of-00189.safetensors", + "model.layers.53.self_attn.k_proj.weight": "model-00107-of-00189.safetensors", + "model.layers.53.self_attn.k_proj.weight_scale": "model-00107-of-00189.safetensors", + "model.layers.53.self_attn.o_proj.weight": "model-00107-of-00189.safetensors", + "model.layers.53.self_attn.o_proj.weight_scale": "model-00107-of-00189.safetensors", + "model.layers.53.self_attn.q_proj.weight": "model-00107-of-00189.safetensors", + "model.layers.53.self_attn.q_proj.weight_scale": "model-00107-of-00189.safetensors", + "model.layers.53.self_attn.v_proj.weight": "model-00107-of-00189.safetensors", + "model.layers.53.self_attn.v_proj.weight_scale": "model-00107-of-00189.safetensors", + "model.layers.54.input_layernorm.bias": "model-00111-of-00189.safetensors", + "model.layers.54.input_layernorm.weight": "model-00111-of-00189.safetensors", + "model.layers.54.mlp.down_proj.weight": "model-00111-of-00189.safetensors", + "model.layers.54.mlp.down_proj.weight_scale": "model-00111-of-00189.safetensors", + "model.layers.54.mlp.up_proj.weight": "model-00110-of-00189.safetensors", + "model.layers.54.mlp.up_proj.weight_scale": "model-00110-of-00189.safetensors", + "model.layers.54.post_attention_layernorm.bias": "model-00111-of-00189.safetensors", + "model.layers.54.post_attention_layernorm.weight": "model-00111-of-00189.safetensors", + "model.layers.54.self_attn.k_proj.weight": "model-00109-of-00189.safetensors", + "model.layers.54.self_attn.k_proj.weight_scale": "model-00109-of-00189.safetensors", + "model.layers.54.self_attn.o_proj.weight": "model-00109-of-00189.safetensors", + "model.layers.54.self_attn.o_proj.weight_scale": "model-00109-of-00189.safetensors", + "model.layers.54.self_attn.q_proj.weight": "model-00109-of-00189.safetensors", + "model.layers.54.self_attn.q_proj.weight_scale": "model-00109-of-00189.safetensors", + "model.layers.54.self_attn.v_proj.weight": "model-00109-of-00189.safetensors", + "model.layers.54.self_attn.v_proj.weight_scale": "model-00109-of-00189.safetensors", + "model.layers.55.input_layernorm.bias": "model-00113-of-00189.safetensors", + "model.layers.55.input_layernorm.weight": "model-00113-of-00189.safetensors", + "model.layers.55.mlp.down_proj.weight": "model-00113-of-00189.safetensors", + "model.layers.55.mlp.down_proj.weight_scale": "model-00113-of-00189.safetensors", + "model.layers.55.mlp.up_proj.weight": "model-00112-of-00189.safetensors", + "model.layers.55.mlp.up_proj.weight_scale": "model-00112-of-00189.safetensors", + "model.layers.55.post_attention_layernorm.bias": "model-00113-of-00189.safetensors", + "model.layers.55.post_attention_layernorm.weight": "model-00113-of-00189.safetensors", + "model.layers.55.self_attn.k_proj.weight": "model-00111-of-00189.safetensors", + "model.layers.55.self_attn.k_proj.weight_scale": "model-00111-of-00189.safetensors", + "model.layers.55.self_attn.o_proj.weight": "model-00111-of-00189.safetensors", + "model.layers.55.self_attn.o_proj.weight_scale": "model-00111-of-00189.safetensors", + "model.layers.55.self_attn.q_proj.weight": "model-00111-of-00189.safetensors", + "model.layers.55.self_attn.q_proj.weight_scale": "model-00111-of-00189.safetensors", + "model.layers.55.self_attn.v_proj.weight": "model-00111-of-00189.safetensors", + "model.layers.55.self_attn.v_proj.weight_scale": "model-00111-of-00189.safetensors", + "model.layers.56.input_layernorm.bias": "model-00115-of-00189.safetensors", + "model.layers.56.input_layernorm.weight": "model-00115-of-00189.safetensors", + "model.layers.56.mlp.down_proj.weight": "model-00115-of-00189.safetensors", + "model.layers.56.mlp.down_proj.weight_scale": "model-00115-of-00189.safetensors", + "model.layers.56.mlp.up_proj.weight": "model-00114-of-00189.safetensors", + "model.layers.56.mlp.up_proj.weight_scale": "model-00114-of-00189.safetensors", + "model.layers.56.post_attention_layernorm.bias": "model-00115-of-00189.safetensors", + "model.layers.56.post_attention_layernorm.weight": "model-00115-of-00189.safetensors", + "model.layers.56.self_attn.k_proj.weight": "model-00113-of-00189.safetensors", + "model.layers.56.self_attn.k_proj.weight_scale": "model-00113-of-00189.safetensors", + "model.layers.56.self_attn.o_proj.weight": "model-00113-of-00189.safetensors", + "model.layers.56.self_attn.o_proj.weight_scale": "model-00113-of-00189.safetensors", + "model.layers.56.self_attn.q_proj.weight": "model-00113-of-00189.safetensors", + "model.layers.56.self_attn.q_proj.weight_scale": "model-00113-of-00189.safetensors", + "model.layers.56.self_attn.v_proj.weight": "model-00113-of-00189.safetensors", + "model.layers.56.self_attn.v_proj.weight_scale": "model-00113-of-00189.safetensors", + "model.layers.57.input_layernorm.bias": "model-00117-of-00189.safetensors", + "model.layers.57.input_layernorm.weight": "model-00117-of-00189.safetensors", + "model.layers.57.mlp.down_proj.weight": "model-00117-of-00189.safetensors", + "model.layers.57.mlp.down_proj.weight_scale": "model-00117-of-00189.safetensors", + "model.layers.57.mlp.up_proj.weight": "model-00116-of-00189.safetensors", + "model.layers.57.mlp.up_proj.weight_scale": "model-00116-of-00189.safetensors", + "model.layers.57.post_attention_layernorm.bias": "model-00117-of-00189.safetensors", + "model.layers.57.post_attention_layernorm.weight": "model-00117-of-00189.safetensors", + "model.layers.57.self_attn.k_proj.weight": "model-00115-of-00189.safetensors", + "model.layers.57.self_attn.k_proj.weight_scale": "model-00115-of-00189.safetensors", + "model.layers.57.self_attn.o_proj.weight": "model-00115-of-00189.safetensors", + "model.layers.57.self_attn.o_proj.weight_scale": "model-00115-of-00189.safetensors", + "model.layers.57.self_attn.q_proj.weight": "model-00115-of-00189.safetensors", + "model.layers.57.self_attn.q_proj.weight_scale": "model-00115-of-00189.safetensors", + "model.layers.57.self_attn.v_proj.weight": "model-00115-of-00189.safetensors", + "model.layers.57.self_attn.v_proj.weight_scale": "model-00115-of-00189.safetensors", + "model.layers.58.input_layernorm.bias": "model-00119-of-00189.safetensors", + "model.layers.58.input_layernorm.weight": "model-00119-of-00189.safetensors", + "model.layers.58.mlp.down_proj.weight": "model-00119-of-00189.safetensors", + "model.layers.58.mlp.down_proj.weight_scale": "model-00119-of-00189.safetensors", + "model.layers.58.mlp.up_proj.weight": "model-00118-of-00189.safetensors", + "model.layers.58.mlp.up_proj.weight_scale": "model-00118-of-00189.safetensors", + "model.layers.58.post_attention_layernorm.bias": "model-00119-of-00189.safetensors", + "model.layers.58.post_attention_layernorm.weight": "model-00119-of-00189.safetensors", + "model.layers.58.self_attn.k_proj.weight": "model-00117-of-00189.safetensors", + "model.layers.58.self_attn.k_proj.weight_scale": "model-00117-of-00189.safetensors", + "model.layers.58.self_attn.o_proj.weight": "model-00117-of-00189.safetensors", + "model.layers.58.self_attn.o_proj.weight_scale": "model-00117-of-00189.safetensors", + "model.layers.58.self_attn.q_proj.weight": "model-00117-of-00189.safetensors", + "model.layers.58.self_attn.q_proj.weight_scale": "model-00117-of-00189.safetensors", + "model.layers.58.self_attn.v_proj.weight": "model-00117-of-00189.safetensors", + "model.layers.58.self_attn.v_proj.weight_scale": "model-00117-of-00189.safetensors", + "model.layers.59.input_layernorm.bias": "model-00121-of-00189.safetensors", + "model.layers.59.input_layernorm.weight": "model-00121-of-00189.safetensors", + "model.layers.59.mlp.down_proj.weight": "model-00121-of-00189.safetensors", + "model.layers.59.mlp.down_proj.weight_scale": "model-00121-of-00189.safetensors", + "model.layers.59.mlp.up_proj.weight": "model-00120-of-00189.safetensors", + "model.layers.59.mlp.up_proj.weight_scale": "model-00120-of-00189.safetensors", + "model.layers.59.post_attention_layernorm.bias": "model-00121-of-00189.safetensors", + "model.layers.59.post_attention_layernorm.weight": "model-00121-of-00189.safetensors", + "model.layers.59.self_attn.k_proj.weight": "model-00119-of-00189.safetensors", + "model.layers.59.self_attn.k_proj.weight_scale": "model-00119-of-00189.safetensors", + "model.layers.59.self_attn.o_proj.weight": "model-00119-of-00189.safetensors", + "model.layers.59.self_attn.o_proj.weight_scale": "model-00119-of-00189.safetensors", + "model.layers.59.self_attn.q_proj.weight": "model-00119-of-00189.safetensors", + "model.layers.59.self_attn.q_proj.weight_scale": "model-00119-of-00189.safetensors", + "model.layers.59.self_attn.v_proj.weight": "model-00119-of-00189.safetensors", + "model.layers.59.self_attn.v_proj.weight_scale": "model-00119-of-00189.safetensors", + "model.layers.6.input_layernorm.bias": "model-00015-of-00189.safetensors", + "model.layers.6.input_layernorm.weight": "model-00015-of-00189.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00015-of-00189.safetensors", + "model.layers.6.mlp.down_proj.weight_scale": "model-00015-of-00189.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00014-of-00189.safetensors", + "model.layers.6.mlp.up_proj.weight_scale": "model-00014-of-00189.safetensors", + "model.layers.6.post_attention_layernorm.bias": "model-00015-of-00189.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00015-of-00189.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00013-of-00189.safetensors", + "model.layers.6.self_attn.k_proj.weight_scale": "model-00013-of-00189.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00013-of-00189.safetensors", + "model.layers.6.self_attn.o_proj.weight_scale": "model-00013-of-00189.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00013-of-00189.safetensors", + "model.layers.6.self_attn.q_proj.weight_scale": "model-00013-of-00189.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00013-of-00189.safetensors", + "model.layers.6.self_attn.v_proj.weight_scale": "model-00013-of-00189.safetensors", + "model.layers.60.input_layernorm.bias": "model-00123-of-00189.safetensors", + "model.layers.60.input_layernorm.weight": "model-00123-of-00189.safetensors", + "model.layers.60.mlp.down_proj.weight": "model-00123-of-00189.safetensors", + "model.layers.60.mlp.down_proj.weight_scale": "model-00123-of-00189.safetensors", + "model.layers.60.mlp.up_proj.weight": "model-00122-of-00189.safetensors", + "model.layers.60.mlp.up_proj.weight_scale": "model-00122-of-00189.safetensors", + "model.layers.60.post_attention_layernorm.bias": "model-00123-of-00189.safetensors", + "model.layers.60.post_attention_layernorm.weight": "model-00123-of-00189.safetensors", + "model.layers.60.self_attn.k_proj.weight": "model-00121-of-00189.safetensors", + "model.layers.60.self_attn.k_proj.weight_scale": "model-00121-of-00189.safetensors", + "model.layers.60.self_attn.o_proj.weight": "model-00121-of-00189.safetensors", + "model.layers.60.self_attn.o_proj.weight_scale": "model-00121-of-00189.safetensors", + "model.layers.60.self_attn.q_proj.weight": "model-00121-of-00189.safetensors", + "model.layers.60.self_attn.q_proj.weight_scale": "model-00121-of-00189.safetensors", + "model.layers.60.self_attn.v_proj.weight": "model-00121-of-00189.safetensors", + "model.layers.60.self_attn.v_proj.weight_scale": "model-00121-of-00189.safetensors", + "model.layers.61.input_layernorm.bias": "model-00125-of-00189.safetensors", + "model.layers.61.input_layernorm.weight": "model-00125-of-00189.safetensors", + "model.layers.61.mlp.down_proj.weight": "model-00125-of-00189.safetensors", + "model.layers.61.mlp.down_proj.weight_scale": "model-00125-of-00189.safetensors", + "model.layers.61.mlp.up_proj.weight": "model-00124-of-00189.safetensors", + "model.layers.61.mlp.up_proj.weight_scale": "model-00124-of-00189.safetensors", + "model.layers.61.post_attention_layernorm.bias": "model-00125-of-00189.safetensors", + "model.layers.61.post_attention_layernorm.weight": "model-00125-of-00189.safetensors", + "model.layers.61.self_attn.k_proj.weight": "model-00123-of-00189.safetensors", + "model.layers.61.self_attn.k_proj.weight_scale": "model-00123-of-00189.safetensors", + "model.layers.61.self_attn.o_proj.weight": "model-00123-of-00189.safetensors", + "model.layers.61.self_attn.o_proj.weight_scale": "model-00123-of-00189.safetensors", + "model.layers.61.self_attn.q_proj.weight": "model-00123-of-00189.safetensors", + "model.layers.61.self_attn.q_proj.weight_scale": "model-00123-of-00189.safetensors", + "model.layers.61.self_attn.v_proj.weight": "model-00123-of-00189.safetensors", + "model.layers.61.self_attn.v_proj.weight_scale": "model-00123-of-00189.safetensors", + "model.layers.62.input_layernorm.bias": "model-00127-of-00189.safetensors", + "model.layers.62.input_layernorm.weight": "model-00127-of-00189.safetensors", + "model.layers.62.mlp.down_proj.weight": "model-00127-of-00189.safetensors", + "model.layers.62.mlp.down_proj.weight_scale": "model-00127-of-00189.safetensors", + "model.layers.62.mlp.up_proj.weight": "model-00126-of-00189.safetensors", + "model.layers.62.mlp.up_proj.weight_scale": "model-00126-of-00189.safetensors", + "model.layers.62.post_attention_layernorm.bias": "model-00127-of-00189.safetensors", + "model.layers.62.post_attention_layernorm.weight": "model-00127-of-00189.safetensors", + "model.layers.62.self_attn.k_proj.weight": "model-00125-of-00189.safetensors", + "model.layers.62.self_attn.k_proj.weight_scale": "model-00125-of-00189.safetensors", + "model.layers.62.self_attn.o_proj.weight": "model-00125-of-00189.safetensors", + "model.layers.62.self_attn.o_proj.weight_scale": "model-00125-of-00189.safetensors", + "model.layers.62.self_attn.q_proj.weight": "model-00125-of-00189.safetensors", + "model.layers.62.self_attn.q_proj.weight_scale": "model-00125-of-00189.safetensors", + "model.layers.62.self_attn.v_proj.weight": "model-00125-of-00189.safetensors", + "model.layers.62.self_attn.v_proj.weight_scale": "model-00125-of-00189.safetensors", + "model.layers.63.input_layernorm.bias": "model-00129-of-00189.safetensors", + "model.layers.63.input_layernorm.weight": "model-00129-of-00189.safetensors", + "model.layers.63.mlp.down_proj.weight": "model-00129-of-00189.safetensors", + "model.layers.63.mlp.down_proj.weight_scale": "model-00129-of-00189.safetensors", + "model.layers.63.mlp.up_proj.weight": "model-00128-of-00189.safetensors", + "model.layers.63.mlp.up_proj.weight_scale": "model-00128-of-00189.safetensors", + "model.layers.63.post_attention_layernorm.bias": "model-00129-of-00189.safetensors", + "model.layers.63.post_attention_layernorm.weight": "model-00129-of-00189.safetensors", + "model.layers.63.self_attn.k_proj.weight": "model-00127-of-00189.safetensors", + "model.layers.63.self_attn.k_proj.weight_scale": "model-00127-of-00189.safetensors", + "model.layers.63.self_attn.o_proj.weight": "model-00127-of-00189.safetensors", + "model.layers.63.self_attn.o_proj.weight_scale": "model-00127-of-00189.safetensors", + "model.layers.63.self_attn.q_proj.weight": "model-00127-of-00189.safetensors", + "model.layers.63.self_attn.q_proj.weight_scale": "model-00127-of-00189.safetensors", + "model.layers.63.self_attn.v_proj.weight": "model-00127-of-00189.safetensors", + "model.layers.63.self_attn.v_proj.weight_scale": "model-00127-of-00189.safetensors", + "model.layers.64.input_layernorm.bias": "model-00131-of-00189.safetensors", + "model.layers.64.input_layernorm.weight": "model-00131-of-00189.safetensors", + "model.layers.64.mlp.down_proj.weight": "model-00131-of-00189.safetensors", + "model.layers.64.mlp.down_proj.weight_scale": "model-00131-of-00189.safetensors", + "model.layers.64.mlp.up_proj.weight": "model-00130-of-00189.safetensors", + "model.layers.64.mlp.up_proj.weight_scale": "model-00130-of-00189.safetensors", + "model.layers.64.post_attention_layernorm.bias": "model-00131-of-00189.safetensors", + "model.layers.64.post_attention_layernorm.weight": "model-00131-of-00189.safetensors", + "model.layers.64.self_attn.k_proj.weight": "model-00129-of-00189.safetensors", + "model.layers.64.self_attn.k_proj.weight_scale": "model-00129-of-00189.safetensors", + "model.layers.64.self_attn.o_proj.weight": "model-00129-of-00189.safetensors", + "model.layers.64.self_attn.o_proj.weight_scale": "model-00129-of-00189.safetensors", + "model.layers.64.self_attn.q_proj.weight": "model-00129-of-00189.safetensors", + "model.layers.64.self_attn.q_proj.weight_scale": "model-00129-of-00189.safetensors", + "model.layers.64.self_attn.v_proj.weight": "model-00129-of-00189.safetensors", + "model.layers.64.self_attn.v_proj.weight_scale": "model-00129-of-00189.safetensors", + "model.layers.65.input_layernorm.bias": "model-00133-of-00189.safetensors", + "model.layers.65.input_layernorm.weight": "model-00133-of-00189.safetensors", + "model.layers.65.mlp.down_proj.weight": "model-00133-of-00189.safetensors", + "model.layers.65.mlp.down_proj.weight_scale": "model-00133-of-00189.safetensors", + "model.layers.65.mlp.up_proj.weight": "model-00132-of-00189.safetensors", + "model.layers.65.mlp.up_proj.weight_scale": "model-00132-of-00189.safetensors", + "model.layers.65.post_attention_layernorm.bias": "model-00133-of-00189.safetensors", + "model.layers.65.post_attention_layernorm.weight": "model-00133-of-00189.safetensors", + "model.layers.65.self_attn.k_proj.weight": "model-00131-of-00189.safetensors", + "model.layers.65.self_attn.k_proj.weight_scale": "model-00131-of-00189.safetensors", + "model.layers.65.self_attn.o_proj.weight": "model-00131-of-00189.safetensors", + "model.layers.65.self_attn.o_proj.weight_scale": "model-00131-of-00189.safetensors", + "model.layers.65.self_attn.q_proj.weight": "model-00131-of-00189.safetensors", + "model.layers.65.self_attn.q_proj.weight_scale": "model-00131-of-00189.safetensors", + "model.layers.65.self_attn.v_proj.weight": "model-00131-of-00189.safetensors", + "model.layers.65.self_attn.v_proj.weight_scale": "model-00131-of-00189.safetensors", + "model.layers.66.input_layernorm.bias": "model-00135-of-00189.safetensors", + "model.layers.66.input_layernorm.weight": "model-00135-of-00189.safetensors", + "model.layers.66.mlp.down_proj.weight": "model-00135-of-00189.safetensors", + "model.layers.66.mlp.down_proj.weight_scale": "model-00135-of-00189.safetensors", + "model.layers.66.mlp.up_proj.weight": "model-00134-of-00189.safetensors", + "model.layers.66.mlp.up_proj.weight_scale": "model-00134-of-00189.safetensors", + "model.layers.66.post_attention_layernorm.bias": "model-00135-of-00189.safetensors", + "model.layers.66.post_attention_layernorm.weight": "model-00135-of-00189.safetensors", + "model.layers.66.self_attn.k_proj.weight": "model-00133-of-00189.safetensors", + "model.layers.66.self_attn.k_proj.weight_scale": "model-00133-of-00189.safetensors", + "model.layers.66.self_attn.o_proj.weight": "model-00133-of-00189.safetensors", + "model.layers.66.self_attn.o_proj.weight_scale": "model-00133-of-00189.safetensors", + "model.layers.66.self_attn.q_proj.weight": "model-00133-of-00189.safetensors", + "model.layers.66.self_attn.q_proj.weight_scale": "model-00133-of-00189.safetensors", + "model.layers.66.self_attn.v_proj.weight": "model-00133-of-00189.safetensors", + "model.layers.66.self_attn.v_proj.weight_scale": "model-00133-of-00189.safetensors", + "model.layers.67.input_layernorm.bias": "model-00137-of-00189.safetensors", + "model.layers.67.input_layernorm.weight": "model-00137-of-00189.safetensors", + "model.layers.67.mlp.down_proj.weight": "model-00137-of-00189.safetensors", + "model.layers.67.mlp.down_proj.weight_scale": "model-00137-of-00189.safetensors", + "model.layers.67.mlp.up_proj.weight": "model-00136-of-00189.safetensors", + "model.layers.67.mlp.up_proj.weight_scale": "model-00136-of-00189.safetensors", + "model.layers.67.post_attention_layernorm.bias": "model-00137-of-00189.safetensors", + "model.layers.67.post_attention_layernorm.weight": "model-00137-of-00189.safetensors", + "model.layers.67.self_attn.k_proj.weight": "model-00135-of-00189.safetensors", + "model.layers.67.self_attn.k_proj.weight_scale": "model-00135-of-00189.safetensors", + "model.layers.67.self_attn.o_proj.weight": "model-00135-of-00189.safetensors", + "model.layers.67.self_attn.o_proj.weight_scale": "model-00135-of-00189.safetensors", + "model.layers.67.self_attn.q_proj.weight": "model-00135-of-00189.safetensors", + "model.layers.67.self_attn.q_proj.weight_scale": "model-00135-of-00189.safetensors", + "model.layers.67.self_attn.v_proj.weight": "model-00135-of-00189.safetensors", + "model.layers.67.self_attn.v_proj.weight_scale": "model-00135-of-00189.safetensors", + "model.layers.68.input_layernorm.bias": "model-00139-of-00189.safetensors", + "model.layers.68.input_layernorm.weight": "model-00139-of-00189.safetensors", + "model.layers.68.mlp.down_proj.weight": "model-00139-of-00189.safetensors", + "model.layers.68.mlp.down_proj.weight_scale": "model-00139-of-00189.safetensors", + "model.layers.68.mlp.up_proj.weight": "model-00138-of-00189.safetensors", + "model.layers.68.mlp.up_proj.weight_scale": "model-00138-of-00189.safetensors", + "model.layers.68.post_attention_layernorm.bias": "model-00139-of-00189.safetensors", + "model.layers.68.post_attention_layernorm.weight": "model-00139-of-00189.safetensors", + "model.layers.68.self_attn.k_proj.weight": "model-00137-of-00189.safetensors", + "model.layers.68.self_attn.k_proj.weight_scale": "model-00137-of-00189.safetensors", + "model.layers.68.self_attn.o_proj.weight": "model-00137-of-00189.safetensors", + "model.layers.68.self_attn.o_proj.weight_scale": "model-00137-of-00189.safetensors", + "model.layers.68.self_attn.q_proj.weight": "model-00137-of-00189.safetensors", + "model.layers.68.self_attn.q_proj.weight_scale": "model-00137-of-00189.safetensors", + "model.layers.68.self_attn.v_proj.weight": "model-00137-of-00189.safetensors", + "model.layers.68.self_attn.v_proj.weight_scale": "model-00137-of-00189.safetensors", + "model.layers.69.input_layernorm.bias": "model-00141-of-00189.safetensors", + "model.layers.69.input_layernorm.weight": "model-00141-of-00189.safetensors", + "model.layers.69.mlp.down_proj.weight": "model-00141-of-00189.safetensors", + "model.layers.69.mlp.down_proj.weight_scale": "model-00141-of-00189.safetensors", + "model.layers.69.mlp.up_proj.weight": "model-00140-of-00189.safetensors", + "model.layers.69.mlp.up_proj.weight_scale": "model-00140-of-00189.safetensors", + "model.layers.69.post_attention_layernorm.bias": "model-00141-of-00189.safetensors", + "model.layers.69.post_attention_layernorm.weight": "model-00141-of-00189.safetensors", + "model.layers.69.self_attn.k_proj.weight": "model-00139-of-00189.safetensors", + "model.layers.69.self_attn.k_proj.weight_scale": "model-00139-of-00189.safetensors", + "model.layers.69.self_attn.o_proj.weight": "model-00139-of-00189.safetensors", + "model.layers.69.self_attn.o_proj.weight_scale": "model-00139-of-00189.safetensors", + "model.layers.69.self_attn.q_proj.weight": "model-00139-of-00189.safetensors", + "model.layers.69.self_attn.q_proj.weight_scale": "model-00139-of-00189.safetensors", + "model.layers.69.self_attn.v_proj.weight": "model-00139-of-00189.safetensors", + "model.layers.69.self_attn.v_proj.weight_scale": "model-00139-of-00189.safetensors", + "model.layers.7.input_layernorm.bias": "model-00017-of-00189.safetensors", + "model.layers.7.input_layernorm.weight": "model-00017-of-00189.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00017-of-00189.safetensors", + "model.layers.7.mlp.down_proj.weight_scale": "model-00017-of-00189.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00016-of-00189.safetensors", + "model.layers.7.mlp.up_proj.weight_scale": "model-00016-of-00189.safetensors", + "model.layers.7.post_attention_layernorm.bias": "model-00017-of-00189.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00017-of-00189.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00015-of-00189.safetensors", + "model.layers.7.self_attn.k_proj.weight_scale": "model-00015-of-00189.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00015-of-00189.safetensors", + "model.layers.7.self_attn.o_proj.weight_scale": "model-00015-of-00189.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00015-of-00189.safetensors", + "model.layers.7.self_attn.q_proj.weight_scale": "model-00015-of-00189.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00015-of-00189.safetensors", + "model.layers.7.self_attn.v_proj.weight_scale": "model-00015-of-00189.safetensors", + "model.layers.70.input_layernorm.bias": "model-00143-of-00189.safetensors", + "model.layers.70.input_layernorm.weight": "model-00143-of-00189.safetensors", + "model.layers.70.mlp.down_proj.weight": "model-00143-of-00189.safetensors", + "model.layers.70.mlp.down_proj.weight_scale": "model-00143-of-00189.safetensors", + "model.layers.70.mlp.up_proj.weight": "model-00142-of-00189.safetensors", + "model.layers.70.mlp.up_proj.weight_scale": "model-00142-of-00189.safetensors", + "model.layers.70.post_attention_layernorm.bias": "model-00143-of-00189.safetensors", + "model.layers.70.post_attention_layernorm.weight": "model-00143-of-00189.safetensors", + "model.layers.70.self_attn.k_proj.weight": "model-00141-of-00189.safetensors", + "model.layers.70.self_attn.k_proj.weight_scale": "model-00141-of-00189.safetensors", + "model.layers.70.self_attn.o_proj.weight": "model-00141-of-00189.safetensors", + "model.layers.70.self_attn.o_proj.weight_scale": "model-00141-of-00189.safetensors", + "model.layers.70.self_attn.q_proj.weight": "model-00141-of-00189.safetensors", + "model.layers.70.self_attn.q_proj.weight_scale": "model-00141-of-00189.safetensors", + "model.layers.70.self_attn.v_proj.weight": "model-00141-of-00189.safetensors", + "model.layers.70.self_attn.v_proj.weight_scale": "model-00141-of-00189.safetensors", + "model.layers.71.input_layernorm.bias": "model-00145-of-00189.safetensors", + "model.layers.71.input_layernorm.weight": "model-00145-of-00189.safetensors", + "model.layers.71.mlp.down_proj.weight": "model-00145-of-00189.safetensors", + "model.layers.71.mlp.down_proj.weight_scale": "model-00145-of-00189.safetensors", + "model.layers.71.mlp.up_proj.weight": "model-00144-of-00189.safetensors", + "model.layers.71.mlp.up_proj.weight_scale": "model-00144-of-00189.safetensors", + "model.layers.71.post_attention_layernorm.bias": "model-00145-of-00189.safetensors", + "model.layers.71.post_attention_layernorm.weight": "model-00145-of-00189.safetensors", + "model.layers.71.self_attn.k_proj.weight": "model-00143-of-00189.safetensors", + "model.layers.71.self_attn.k_proj.weight_scale": "model-00143-of-00189.safetensors", + "model.layers.71.self_attn.o_proj.weight": "model-00143-of-00189.safetensors", + "model.layers.71.self_attn.o_proj.weight_scale": "model-00143-of-00189.safetensors", + "model.layers.71.self_attn.q_proj.weight": "model-00143-of-00189.safetensors", + "model.layers.71.self_attn.q_proj.weight_scale": "model-00143-of-00189.safetensors", + "model.layers.71.self_attn.v_proj.weight": "model-00143-of-00189.safetensors", + "model.layers.71.self_attn.v_proj.weight_scale": "model-00143-of-00189.safetensors", + "model.layers.72.input_layernorm.bias": "model-00147-of-00189.safetensors", + "model.layers.72.input_layernorm.weight": "model-00147-of-00189.safetensors", + "model.layers.72.mlp.down_proj.weight": "model-00147-of-00189.safetensors", + "model.layers.72.mlp.down_proj.weight_scale": "model-00147-of-00189.safetensors", + "model.layers.72.mlp.up_proj.weight": "model-00146-of-00189.safetensors", + "model.layers.72.mlp.up_proj.weight_scale": "model-00146-of-00189.safetensors", + "model.layers.72.post_attention_layernorm.bias": "model-00147-of-00189.safetensors", + "model.layers.72.post_attention_layernorm.weight": "model-00147-of-00189.safetensors", + "model.layers.72.self_attn.k_proj.weight": "model-00145-of-00189.safetensors", + "model.layers.72.self_attn.k_proj.weight_scale": "model-00145-of-00189.safetensors", + "model.layers.72.self_attn.o_proj.weight": "model-00145-of-00189.safetensors", + "model.layers.72.self_attn.o_proj.weight_scale": "model-00145-of-00189.safetensors", + "model.layers.72.self_attn.q_proj.weight": "model-00145-of-00189.safetensors", + "model.layers.72.self_attn.q_proj.weight_scale": "model-00145-of-00189.safetensors", + "model.layers.72.self_attn.v_proj.weight": "model-00145-of-00189.safetensors", + "model.layers.72.self_attn.v_proj.weight_scale": "model-00145-of-00189.safetensors", + "model.layers.73.input_layernorm.bias": "model-00149-of-00189.safetensors", + "model.layers.73.input_layernorm.weight": "model-00149-of-00189.safetensors", + "model.layers.73.mlp.down_proj.weight": "model-00149-of-00189.safetensors", + "model.layers.73.mlp.down_proj.weight_scale": "model-00149-of-00189.safetensors", + "model.layers.73.mlp.up_proj.weight": "model-00148-of-00189.safetensors", + "model.layers.73.mlp.up_proj.weight_scale": "model-00148-of-00189.safetensors", + "model.layers.73.post_attention_layernorm.bias": "model-00149-of-00189.safetensors", + "model.layers.73.post_attention_layernorm.weight": "model-00149-of-00189.safetensors", + "model.layers.73.self_attn.k_proj.weight": "model-00147-of-00189.safetensors", + "model.layers.73.self_attn.k_proj.weight_scale": "model-00147-of-00189.safetensors", + "model.layers.73.self_attn.o_proj.weight": "model-00147-of-00189.safetensors", + "model.layers.73.self_attn.o_proj.weight_scale": "model-00147-of-00189.safetensors", + "model.layers.73.self_attn.q_proj.weight": "model-00147-of-00189.safetensors", + "model.layers.73.self_attn.q_proj.weight_scale": "model-00147-of-00189.safetensors", + "model.layers.73.self_attn.v_proj.weight": "model-00147-of-00189.safetensors", + "model.layers.73.self_attn.v_proj.weight_scale": "model-00147-of-00189.safetensors", + "model.layers.74.input_layernorm.bias": "model-00151-of-00189.safetensors", + "model.layers.74.input_layernorm.weight": "model-00151-of-00189.safetensors", + "model.layers.74.mlp.down_proj.weight": "model-00151-of-00189.safetensors", + "model.layers.74.mlp.down_proj.weight_scale": "model-00151-of-00189.safetensors", + "model.layers.74.mlp.up_proj.weight": "model-00150-of-00189.safetensors", + "model.layers.74.mlp.up_proj.weight_scale": "model-00150-of-00189.safetensors", + "model.layers.74.post_attention_layernorm.bias": "model-00151-of-00189.safetensors", + "model.layers.74.post_attention_layernorm.weight": "model-00151-of-00189.safetensors", + "model.layers.74.self_attn.k_proj.weight": "model-00149-of-00189.safetensors", + "model.layers.74.self_attn.k_proj.weight_scale": "model-00149-of-00189.safetensors", + "model.layers.74.self_attn.o_proj.weight": "model-00149-of-00189.safetensors", + "model.layers.74.self_attn.o_proj.weight_scale": "model-00149-of-00189.safetensors", + "model.layers.74.self_attn.q_proj.weight": "model-00149-of-00189.safetensors", + "model.layers.74.self_attn.q_proj.weight_scale": "model-00149-of-00189.safetensors", + "model.layers.74.self_attn.v_proj.weight": "model-00149-of-00189.safetensors", + "model.layers.74.self_attn.v_proj.weight_scale": "model-00149-of-00189.safetensors", + "model.layers.75.input_layernorm.bias": "model-00153-of-00189.safetensors", + "model.layers.75.input_layernorm.weight": "model-00153-of-00189.safetensors", + "model.layers.75.mlp.down_proj.weight": "model-00153-of-00189.safetensors", + "model.layers.75.mlp.down_proj.weight_scale": "model-00153-of-00189.safetensors", + "model.layers.75.mlp.up_proj.weight": "model-00152-of-00189.safetensors", + "model.layers.75.mlp.up_proj.weight_scale": "model-00152-of-00189.safetensors", + "model.layers.75.post_attention_layernorm.bias": "model-00153-of-00189.safetensors", + "model.layers.75.post_attention_layernorm.weight": "model-00153-of-00189.safetensors", + "model.layers.75.self_attn.k_proj.weight": "model-00151-of-00189.safetensors", + "model.layers.75.self_attn.k_proj.weight_scale": "model-00151-of-00189.safetensors", + "model.layers.75.self_attn.o_proj.weight": "model-00151-of-00189.safetensors", + "model.layers.75.self_attn.o_proj.weight_scale": "model-00151-of-00189.safetensors", + "model.layers.75.self_attn.q_proj.weight": "model-00151-of-00189.safetensors", + "model.layers.75.self_attn.q_proj.weight_scale": "model-00151-of-00189.safetensors", + "model.layers.75.self_attn.v_proj.weight": "model-00151-of-00189.safetensors", + "model.layers.75.self_attn.v_proj.weight_scale": "model-00151-of-00189.safetensors", + "model.layers.76.input_layernorm.bias": "model-00155-of-00189.safetensors", + "model.layers.76.input_layernorm.weight": "model-00155-of-00189.safetensors", + "model.layers.76.mlp.down_proj.weight": "model-00155-of-00189.safetensors", + "model.layers.76.mlp.down_proj.weight_scale": "model-00155-of-00189.safetensors", + "model.layers.76.mlp.up_proj.weight": "model-00154-of-00189.safetensors", + "model.layers.76.mlp.up_proj.weight_scale": "model-00154-of-00189.safetensors", + "model.layers.76.post_attention_layernorm.bias": "model-00155-of-00189.safetensors", + "model.layers.76.post_attention_layernorm.weight": "model-00155-of-00189.safetensors", + "model.layers.76.self_attn.k_proj.weight": "model-00153-of-00189.safetensors", + "model.layers.76.self_attn.k_proj.weight_scale": "model-00153-of-00189.safetensors", + "model.layers.76.self_attn.o_proj.weight": "model-00153-of-00189.safetensors", + "model.layers.76.self_attn.o_proj.weight_scale": "model-00153-of-00189.safetensors", + "model.layers.76.self_attn.q_proj.weight": "model-00153-of-00189.safetensors", + "model.layers.76.self_attn.q_proj.weight_scale": "model-00153-of-00189.safetensors", + "model.layers.76.self_attn.v_proj.weight": "model-00153-of-00189.safetensors", + "model.layers.76.self_attn.v_proj.weight_scale": "model-00153-of-00189.safetensors", + "model.layers.77.input_layernorm.bias": "model-00157-of-00189.safetensors", + "model.layers.77.input_layernorm.weight": "model-00157-of-00189.safetensors", + "model.layers.77.mlp.down_proj.weight": "model-00157-of-00189.safetensors", + "model.layers.77.mlp.down_proj.weight_scale": "model-00157-of-00189.safetensors", + "model.layers.77.mlp.up_proj.weight": "model-00156-of-00189.safetensors", + "model.layers.77.mlp.up_proj.weight_scale": "model-00156-of-00189.safetensors", + "model.layers.77.post_attention_layernorm.bias": "model-00157-of-00189.safetensors", + "model.layers.77.post_attention_layernorm.weight": "model-00157-of-00189.safetensors", + "model.layers.77.self_attn.k_proj.weight": "model-00155-of-00189.safetensors", + "model.layers.77.self_attn.k_proj.weight_scale": "model-00155-of-00189.safetensors", + "model.layers.77.self_attn.o_proj.weight": "model-00155-of-00189.safetensors", + "model.layers.77.self_attn.o_proj.weight_scale": "model-00155-of-00189.safetensors", + "model.layers.77.self_attn.q_proj.weight": "model-00155-of-00189.safetensors", + "model.layers.77.self_attn.q_proj.weight_scale": "model-00155-of-00189.safetensors", + "model.layers.77.self_attn.v_proj.weight": "model-00155-of-00189.safetensors", + "model.layers.77.self_attn.v_proj.weight_scale": "model-00155-of-00189.safetensors", + "model.layers.78.input_layernorm.bias": "model-00159-of-00189.safetensors", + "model.layers.78.input_layernorm.weight": "model-00159-of-00189.safetensors", + "model.layers.78.mlp.down_proj.weight": "model-00159-of-00189.safetensors", + "model.layers.78.mlp.down_proj.weight_scale": "model-00159-of-00189.safetensors", + "model.layers.78.mlp.up_proj.weight": "model-00158-of-00189.safetensors", + "model.layers.78.mlp.up_proj.weight_scale": "model-00158-of-00189.safetensors", + "model.layers.78.post_attention_layernorm.bias": "model-00159-of-00189.safetensors", + "model.layers.78.post_attention_layernorm.weight": "model-00159-of-00189.safetensors", + "model.layers.78.self_attn.k_proj.weight": "model-00157-of-00189.safetensors", + "model.layers.78.self_attn.k_proj.weight_scale": "model-00157-of-00189.safetensors", + "model.layers.78.self_attn.o_proj.weight": "model-00157-of-00189.safetensors", + "model.layers.78.self_attn.o_proj.weight_scale": "model-00157-of-00189.safetensors", + "model.layers.78.self_attn.q_proj.weight": "model-00157-of-00189.safetensors", + "model.layers.78.self_attn.q_proj.weight_scale": "model-00157-of-00189.safetensors", + "model.layers.78.self_attn.v_proj.weight": "model-00157-of-00189.safetensors", + "model.layers.78.self_attn.v_proj.weight_scale": "model-00157-of-00189.safetensors", + "model.layers.79.input_layernorm.bias": "model-00161-of-00189.safetensors", + "model.layers.79.input_layernorm.weight": "model-00161-of-00189.safetensors", + "model.layers.79.mlp.down_proj.weight": "model-00161-of-00189.safetensors", + "model.layers.79.mlp.down_proj.weight_scale": "model-00161-of-00189.safetensors", + "model.layers.79.mlp.up_proj.weight": "model-00160-of-00189.safetensors", + "model.layers.79.mlp.up_proj.weight_scale": "model-00160-of-00189.safetensors", + "model.layers.79.post_attention_layernorm.bias": "model-00161-of-00189.safetensors", + "model.layers.79.post_attention_layernorm.weight": "model-00161-of-00189.safetensors", + "model.layers.79.self_attn.k_proj.weight": "model-00159-of-00189.safetensors", + "model.layers.79.self_attn.k_proj.weight_scale": "model-00159-of-00189.safetensors", + "model.layers.79.self_attn.o_proj.weight": "model-00159-of-00189.safetensors", + "model.layers.79.self_attn.o_proj.weight_scale": "model-00159-of-00189.safetensors", + "model.layers.79.self_attn.q_proj.weight": "model-00159-of-00189.safetensors", + "model.layers.79.self_attn.q_proj.weight_scale": "model-00159-of-00189.safetensors", + "model.layers.79.self_attn.v_proj.weight": "model-00159-of-00189.safetensors", + "model.layers.79.self_attn.v_proj.weight_scale": "model-00159-of-00189.safetensors", + "model.layers.8.input_layernorm.bias": "model-00019-of-00189.safetensors", + "model.layers.8.input_layernorm.weight": "model-00019-of-00189.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00019-of-00189.safetensors", + "model.layers.8.mlp.down_proj.weight_scale": "model-00019-of-00189.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00018-of-00189.safetensors", + "model.layers.8.mlp.up_proj.weight_scale": "model-00018-of-00189.safetensors", + "model.layers.8.post_attention_layernorm.bias": "model-00019-of-00189.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00019-of-00189.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00017-of-00189.safetensors", + "model.layers.8.self_attn.k_proj.weight_scale": "model-00017-of-00189.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00017-of-00189.safetensors", + "model.layers.8.self_attn.o_proj.weight_scale": "model-00017-of-00189.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00017-of-00189.safetensors", + "model.layers.8.self_attn.q_proj.weight_scale": "model-00017-of-00189.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00017-of-00189.safetensors", + "model.layers.8.self_attn.v_proj.weight_scale": "model-00017-of-00189.safetensors", + "model.layers.80.input_layernorm.bias": "model-00163-of-00189.safetensors", + "model.layers.80.input_layernorm.weight": "model-00163-of-00189.safetensors", + "model.layers.80.mlp.down_proj.weight": "model-00163-of-00189.safetensors", + "model.layers.80.mlp.down_proj.weight_scale": "model-00163-of-00189.safetensors", + "model.layers.80.mlp.up_proj.weight": "model-00162-of-00189.safetensors", + "model.layers.80.mlp.up_proj.weight_scale": "model-00162-of-00189.safetensors", + "model.layers.80.post_attention_layernorm.bias": "model-00163-of-00189.safetensors", + "model.layers.80.post_attention_layernorm.weight": "model-00163-of-00189.safetensors", + "model.layers.80.self_attn.k_proj.weight": "model-00161-of-00189.safetensors", + "model.layers.80.self_attn.k_proj.weight_scale": "model-00161-of-00189.safetensors", + "model.layers.80.self_attn.o_proj.weight": "model-00161-of-00189.safetensors", + "model.layers.80.self_attn.o_proj.weight_scale": "model-00161-of-00189.safetensors", + "model.layers.80.self_attn.q_proj.weight": "model-00161-of-00189.safetensors", + "model.layers.80.self_attn.q_proj.weight_scale": "model-00161-of-00189.safetensors", + "model.layers.80.self_attn.v_proj.weight": "model-00161-of-00189.safetensors", + "model.layers.80.self_attn.v_proj.weight_scale": "model-00161-of-00189.safetensors", + "model.layers.81.input_layernorm.bias": "model-00165-of-00189.safetensors", + "model.layers.81.input_layernorm.weight": "model-00165-of-00189.safetensors", + "model.layers.81.mlp.down_proj.weight": "model-00165-of-00189.safetensors", + "model.layers.81.mlp.down_proj.weight_scale": "model-00165-of-00189.safetensors", + "model.layers.81.mlp.up_proj.weight": "model-00164-of-00189.safetensors", + "model.layers.81.mlp.up_proj.weight_scale": "model-00164-of-00189.safetensors", + "model.layers.81.post_attention_layernorm.bias": "model-00165-of-00189.safetensors", + "model.layers.81.post_attention_layernorm.weight": "model-00165-of-00189.safetensors", + "model.layers.81.self_attn.k_proj.weight": "model-00163-of-00189.safetensors", + "model.layers.81.self_attn.k_proj.weight_scale": "model-00163-of-00189.safetensors", + "model.layers.81.self_attn.o_proj.weight": "model-00163-of-00189.safetensors", + "model.layers.81.self_attn.o_proj.weight_scale": "model-00163-of-00189.safetensors", + "model.layers.81.self_attn.q_proj.weight": "model-00163-of-00189.safetensors", + "model.layers.81.self_attn.q_proj.weight_scale": "model-00163-of-00189.safetensors", + "model.layers.81.self_attn.v_proj.weight": "model-00163-of-00189.safetensors", + "model.layers.81.self_attn.v_proj.weight_scale": "model-00163-of-00189.safetensors", + "model.layers.82.input_layernorm.bias": "model-00167-of-00189.safetensors", + "model.layers.82.input_layernorm.weight": "model-00167-of-00189.safetensors", + "model.layers.82.mlp.down_proj.weight": "model-00167-of-00189.safetensors", + "model.layers.82.mlp.down_proj.weight_scale": "model-00167-of-00189.safetensors", + "model.layers.82.mlp.up_proj.weight": "model-00166-of-00189.safetensors", + "model.layers.82.mlp.up_proj.weight_scale": "model-00166-of-00189.safetensors", + "model.layers.82.post_attention_layernorm.bias": "model-00167-of-00189.safetensors", + "model.layers.82.post_attention_layernorm.weight": "model-00167-of-00189.safetensors", + "model.layers.82.self_attn.k_proj.weight": "model-00165-of-00189.safetensors", + "model.layers.82.self_attn.k_proj.weight_scale": "model-00165-of-00189.safetensors", + "model.layers.82.self_attn.o_proj.weight": "model-00165-of-00189.safetensors", + "model.layers.82.self_attn.o_proj.weight_scale": "model-00165-of-00189.safetensors", + "model.layers.82.self_attn.q_proj.weight": "model-00165-of-00189.safetensors", + "model.layers.82.self_attn.q_proj.weight_scale": "model-00165-of-00189.safetensors", + "model.layers.82.self_attn.v_proj.weight": "model-00165-of-00189.safetensors", + "model.layers.82.self_attn.v_proj.weight_scale": "model-00165-of-00189.safetensors", + "model.layers.83.input_layernorm.bias": "model-00169-of-00189.safetensors", + "model.layers.83.input_layernorm.weight": "model-00169-of-00189.safetensors", + "model.layers.83.mlp.down_proj.weight": "model-00169-of-00189.safetensors", + "model.layers.83.mlp.down_proj.weight_scale": "model-00169-of-00189.safetensors", + "model.layers.83.mlp.up_proj.weight": "model-00168-of-00189.safetensors", + "model.layers.83.mlp.up_proj.weight_scale": "model-00168-of-00189.safetensors", + "model.layers.83.post_attention_layernorm.bias": "model-00169-of-00189.safetensors", + "model.layers.83.post_attention_layernorm.weight": "model-00169-of-00189.safetensors", + "model.layers.83.self_attn.k_proj.weight": "model-00167-of-00189.safetensors", + "model.layers.83.self_attn.k_proj.weight_scale": "model-00167-of-00189.safetensors", + "model.layers.83.self_attn.o_proj.weight": "model-00167-of-00189.safetensors", + "model.layers.83.self_attn.o_proj.weight_scale": "model-00167-of-00189.safetensors", + "model.layers.83.self_attn.q_proj.weight": "model-00167-of-00189.safetensors", + "model.layers.83.self_attn.q_proj.weight_scale": "model-00167-of-00189.safetensors", + "model.layers.83.self_attn.v_proj.weight": "model-00167-of-00189.safetensors", + "model.layers.83.self_attn.v_proj.weight_scale": "model-00167-of-00189.safetensors", + "model.layers.84.input_layernorm.bias": "model-00171-of-00189.safetensors", + "model.layers.84.input_layernorm.weight": "model-00171-of-00189.safetensors", + "model.layers.84.mlp.down_proj.weight": "model-00171-of-00189.safetensors", + "model.layers.84.mlp.down_proj.weight_scale": "model-00171-of-00189.safetensors", + "model.layers.84.mlp.up_proj.weight": "model-00170-of-00189.safetensors", + "model.layers.84.mlp.up_proj.weight_scale": "model-00170-of-00189.safetensors", + "model.layers.84.post_attention_layernorm.bias": "model-00171-of-00189.safetensors", + "model.layers.84.post_attention_layernorm.weight": "model-00171-of-00189.safetensors", + "model.layers.84.self_attn.k_proj.weight": "model-00169-of-00189.safetensors", + "model.layers.84.self_attn.k_proj.weight_scale": "model-00169-of-00189.safetensors", + "model.layers.84.self_attn.o_proj.weight": "model-00169-of-00189.safetensors", + "model.layers.84.self_attn.o_proj.weight_scale": "model-00169-of-00189.safetensors", + "model.layers.84.self_attn.q_proj.weight": "model-00169-of-00189.safetensors", + "model.layers.84.self_attn.q_proj.weight_scale": "model-00169-of-00189.safetensors", + "model.layers.84.self_attn.v_proj.weight": "model-00169-of-00189.safetensors", + "model.layers.84.self_attn.v_proj.weight_scale": "model-00169-of-00189.safetensors", + "model.layers.85.input_layernorm.bias": "model-00173-of-00189.safetensors", + "model.layers.85.input_layernorm.weight": "model-00173-of-00189.safetensors", + "model.layers.85.mlp.down_proj.weight": "model-00173-of-00189.safetensors", + "model.layers.85.mlp.down_proj.weight_scale": "model-00173-of-00189.safetensors", + "model.layers.85.mlp.up_proj.weight": "model-00172-of-00189.safetensors", + "model.layers.85.mlp.up_proj.weight_scale": "model-00172-of-00189.safetensors", + "model.layers.85.post_attention_layernorm.bias": "model-00173-of-00189.safetensors", + "model.layers.85.post_attention_layernorm.weight": "model-00173-of-00189.safetensors", + "model.layers.85.self_attn.k_proj.weight": "model-00171-of-00189.safetensors", + "model.layers.85.self_attn.k_proj.weight_scale": "model-00171-of-00189.safetensors", + "model.layers.85.self_attn.o_proj.weight": "model-00171-of-00189.safetensors", + "model.layers.85.self_attn.o_proj.weight_scale": "model-00171-of-00189.safetensors", + "model.layers.85.self_attn.q_proj.weight": "model-00171-of-00189.safetensors", + "model.layers.85.self_attn.q_proj.weight_scale": "model-00171-of-00189.safetensors", + "model.layers.85.self_attn.v_proj.weight": "model-00171-of-00189.safetensors", + "model.layers.85.self_attn.v_proj.weight_scale": "model-00171-of-00189.safetensors", + "model.layers.86.input_layernorm.bias": "model-00175-of-00189.safetensors", + "model.layers.86.input_layernorm.weight": "model-00175-of-00189.safetensors", + "model.layers.86.mlp.down_proj.weight": "model-00175-of-00189.safetensors", + "model.layers.86.mlp.down_proj.weight_scale": "model-00175-of-00189.safetensors", + "model.layers.86.mlp.up_proj.weight": "model-00174-of-00189.safetensors", + "model.layers.86.mlp.up_proj.weight_scale": "model-00174-of-00189.safetensors", + "model.layers.86.post_attention_layernorm.bias": "model-00175-of-00189.safetensors", + "model.layers.86.post_attention_layernorm.weight": "model-00175-of-00189.safetensors", + "model.layers.86.self_attn.k_proj.weight": "model-00173-of-00189.safetensors", + "model.layers.86.self_attn.k_proj.weight_scale": "model-00173-of-00189.safetensors", + "model.layers.86.self_attn.o_proj.weight": "model-00173-of-00189.safetensors", + "model.layers.86.self_attn.o_proj.weight_scale": "model-00173-of-00189.safetensors", + "model.layers.86.self_attn.q_proj.weight": "model-00173-of-00189.safetensors", + "model.layers.86.self_attn.q_proj.weight_scale": "model-00173-of-00189.safetensors", + "model.layers.86.self_attn.v_proj.weight": "model-00173-of-00189.safetensors", + "model.layers.86.self_attn.v_proj.weight_scale": "model-00173-of-00189.safetensors", + "model.layers.87.input_layernorm.bias": "model-00177-of-00189.safetensors", + "model.layers.87.input_layernorm.weight": "model-00177-of-00189.safetensors", + "model.layers.87.mlp.down_proj.weight": "model-00177-of-00189.safetensors", + "model.layers.87.mlp.down_proj.weight_scale": "model-00177-of-00189.safetensors", + "model.layers.87.mlp.up_proj.weight": "model-00176-of-00189.safetensors", + "model.layers.87.mlp.up_proj.weight_scale": "model-00176-of-00189.safetensors", + "model.layers.87.post_attention_layernorm.bias": "model-00177-of-00189.safetensors", + "model.layers.87.post_attention_layernorm.weight": "model-00177-of-00189.safetensors", + "model.layers.87.self_attn.k_proj.weight": "model-00175-of-00189.safetensors", + "model.layers.87.self_attn.k_proj.weight_scale": "model-00175-of-00189.safetensors", + "model.layers.87.self_attn.o_proj.weight": "model-00175-of-00189.safetensors", + "model.layers.87.self_attn.o_proj.weight_scale": "model-00175-of-00189.safetensors", + "model.layers.87.self_attn.q_proj.weight": "model-00175-of-00189.safetensors", + "model.layers.87.self_attn.q_proj.weight_scale": "model-00175-of-00189.safetensors", + "model.layers.87.self_attn.v_proj.weight": "model-00175-of-00189.safetensors", + "model.layers.87.self_attn.v_proj.weight_scale": "model-00175-of-00189.safetensors", + "model.layers.88.input_layernorm.bias": "model-00179-of-00189.safetensors", + "model.layers.88.input_layernorm.weight": "model-00179-of-00189.safetensors", + "model.layers.88.mlp.down_proj.weight": "model-00179-of-00189.safetensors", + "model.layers.88.mlp.down_proj.weight_scale": "model-00179-of-00189.safetensors", + "model.layers.88.mlp.up_proj.weight": "model-00178-of-00189.safetensors", + "model.layers.88.mlp.up_proj.weight_scale": "model-00178-of-00189.safetensors", + "model.layers.88.post_attention_layernorm.bias": "model-00179-of-00189.safetensors", + "model.layers.88.post_attention_layernorm.weight": "model-00179-of-00189.safetensors", + "model.layers.88.self_attn.k_proj.weight": "model-00177-of-00189.safetensors", + "model.layers.88.self_attn.k_proj.weight_scale": "model-00177-of-00189.safetensors", + "model.layers.88.self_attn.o_proj.weight": "model-00177-of-00189.safetensors", + "model.layers.88.self_attn.o_proj.weight_scale": "model-00177-of-00189.safetensors", + "model.layers.88.self_attn.q_proj.weight": "model-00177-of-00189.safetensors", + "model.layers.88.self_attn.q_proj.weight_scale": "model-00177-of-00189.safetensors", + "model.layers.88.self_attn.v_proj.weight": "model-00177-of-00189.safetensors", + "model.layers.88.self_attn.v_proj.weight_scale": "model-00177-of-00189.safetensors", + "model.layers.89.input_layernorm.bias": "model-00181-of-00189.safetensors", + "model.layers.89.input_layernorm.weight": "model-00181-of-00189.safetensors", + "model.layers.89.mlp.down_proj.weight": "model-00181-of-00189.safetensors", + "model.layers.89.mlp.down_proj.weight_scale": "model-00181-of-00189.safetensors", + "model.layers.89.mlp.up_proj.weight": "model-00180-of-00189.safetensors", + "model.layers.89.mlp.up_proj.weight_scale": "model-00180-of-00189.safetensors", + "model.layers.89.post_attention_layernorm.bias": "model-00181-of-00189.safetensors", + "model.layers.89.post_attention_layernorm.weight": "model-00181-of-00189.safetensors", + "model.layers.89.self_attn.k_proj.weight": "model-00179-of-00189.safetensors", + "model.layers.89.self_attn.k_proj.weight_scale": "model-00179-of-00189.safetensors", + "model.layers.89.self_attn.o_proj.weight": "model-00179-of-00189.safetensors", + "model.layers.89.self_attn.o_proj.weight_scale": "model-00179-of-00189.safetensors", + "model.layers.89.self_attn.q_proj.weight": "model-00179-of-00189.safetensors", + "model.layers.89.self_attn.q_proj.weight_scale": "model-00179-of-00189.safetensors", + "model.layers.89.self_attn.v_proj.weight": "model-00179-of-00189.safetensors", + "model.layers.89.self_attn.v_proj.weight_scale": "model-00179-of-00189.safetensors", + "model.layers.9.input_layernorm.bias": "model-00021-of-00189.safetensors", + "model.layers.9.input_layernorm.weight": "model-00021-of-00189.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00021-of-00189.safetensors", + "model.layers.9.mlp.down_proj.weight_scale": "model-00021-of-00189.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00020-of-00189.safetensors", + "model.layers.9.mlp.up_proj.weight_scale": "model-00020-of-00189.safetensors", + "model.layers.9.post_attention_layernorm.bias": "model-00021-of-00189.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00021-of-00189.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00019-of-00189.safetensors", + "model.layers.9.self_attn.k_proj.weight_scale": "model-00019-of-00189.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00019-of-00189.safetensors", + "model.layers.9.self_attn.o_proj.weight_scale": "model-00019-of-00189.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00019-of-00189.safetensors", + "model.layers.9.self_attn.q_proj.weight_scale": "model-00019-of-00189.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00019-of-00189.safetensors", + "model.layers.9.self_attn.v_proj.weight_scale": "model-00019-of-00189.safetensors", + "model.layers.90.input_layernorm.bias": "model-00183-of-00189.safetensors", + "model.layers.90.input_layernorm.weight": "model-00183-of-00189.safetensors", + "model.layers.90.mlp.down_proj.weight": "model-00183-of-00189.safetensors", + "model.layers.90.mlp.down_proj.weight_scale": "model-00183-of-00189.safetensors", + "model.layers.90.mlp.up_proj.weight": "model-00182-of-00189.safetensors", + "model.layers.90.mlp.up_proj.weight_scale": "model-00182-of-00189.safetensors", + "model.layers.90.post_attention_layernorm.bias": "model-00183-of-00189.safetensors", + "model.layers.90.post_attention_layernorm.weight": "model-00183-of-00189.safetensors", + "model.layers.90.self_attn.k_proj.weight": "model-00181-of-00189.safetensors", + "model.layers.90.self_attn.k_proj.weight_scale": "model-00181-of-00189.safetensors", + "model.layers.90.self_attn.o_proj.weight": "model-00181-of-00189.safetensors", + "model.layers.90.self_attn.o_proj.weight_scale": "model-00181-of-00189.safetensors", + "model.layers.90.self_attn.q_proj.weight": "model-00181-of-00189.safetensors", + "model.layers.90.self_attn.q_proj.weight_scale": "model-00181-of-00189.safetensors", + "model.layers.90.self_attn.v_proj.weight": "model-00181-of-00189.safetensors", + "model.layers.90.self_attn.v_proj.weight_scale": "model-00181-of-00189.safetensors", + "model.layers.91.input_layernorm.bias": "model-00185-of-00189.safetensors", + "model.layers.91.input_layernorm.weight": "model-00185-of-00189.safetensors", + "model.layers.91.mlp.down_proj.weight": "model-00185-of-00189.safetensors", + "model.layers.91.mlp.down_proj.weight_scale": "model-00185-of-00189.safetensors", + "model.layers.91.mlp.up_proj.weight": "model-00184-of-00189.safetensors", + "model.layers.91.mlp.up_proj.weight_scale": "model-00184-of-00189.safetensors", + "model.layers.91.post_attention_layernorm.bias": "model-00185-of-00189.safetensors", + "model.layers.91.post_attention_layernorm.weight": "model-00185-of-00189.safetensors", + "model.layers.91.self_attn.k_proj.weight": "model-00183-of-00189.safetensors", + "model.layers.91.self_attn.k_proj.weight_scale": "model-00183-of-00189.safetensors", + "model.layers.91.self_attn.o_proj.weight": "model-00183-of-00189.safetensors", + "model.layers.91.self_attn.o_proj.weight_scale": "model-00183-of-00189.safetensors", + "model.layers.91.self_attn.q_proj.weight": "model-00183-of-00189.safetensors", + "model.layers.91.self_attn.q_proj.weight_scale": "model-00183-of-00189.safetensors", + "model.layers.91.self_attn.v_proj.weight": "model-00183-of-00189.safetensors", + "model.layers.91.self_attn.v_proj.weight_scale": "model-00183-of-00189.safetensors", + "model.layers.92.input_layernorm.bias": "model-00187-of-00189.safetensors", + "model.layers.92.input_layernorm.weight": "model-00187-of-00189.safetensors", + "model.layers.92.mlp.down_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.92.mlp.down_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.92.mlp.up_proj.weight": "model-00186-of-00189.safetensors", + "model.layers.92.mlp.up_proj.weight_scale": "model-00186-of-00189.safetensors", + "model.layers.92.post_attention_layernorm.bias": "model-00187-of-00189.safetensors", + "model.layers.92.post_attention_layernorm.weight": "model-00187-of-00189.safetensors", + "model.layers.92.self_attn.k_proj.weight": "model-00185-of-00189.safetensors", + "model.layers.92.self_attn.k_proj.weight_scale": "model-00185-of-00189.safetensors", + "model.layers.92.self_attn.o_proj.weight": "model-00185-of-00189.safetensors", + "model.layers.92.self_attn.o_proj.weight_scale": "model-00185-of-00189.safetensors", + "model.layers.92.self_attn.q_proj.weight": "model-00185-of-00189.safetensors", + "model.layers.92.self_attn.q_proj.weight_scale": "model-00185-of-00189.safetensors", + "model.layers.92.self_attn.v_proj.weight": "model-00185-of-00189.safetensors", + "model.layers.92.self_attn.v_proj.weight_scale": "model-00185-of-00189.safetensors", + "model.layers.93.input_layernorm.bias": "model-00189-of-00189.safetensors", + "model.layers.93.input_layernorm.weight": "model-00189-of-00189.safetensors", + "model.layers.93.mlp.down_proj.weight": "model-00189-of-00189.safetensors", + "model.layers.93.mlp.down_proj.weight_scale": "model-00189-of-00189.safetensors", + "model.layers.93.mlp.up_proj.weight": "model-00189-of-00189.safetensors", + "model.layers.93.mlp.up_proj.weight_scale": "model-00189-of-00189.safetensors", + "model.layers.93.post_attention_layernorm.bias": "model-00189-of-00189.safetensors", + "model.layers.93.post_attention_layernorm.weight": "model-00189-of-00189.safetensors", + "model.layers.93.self_attn.k_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.93.self_attn.k_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.93.self_attn.o_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.93.self_attn.o_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.93.self_attn.q_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.93.self_attn.q_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.93.self_attn.v_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.93.self_attn.v_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.94.input_layernorm.bias": "model-00189-of-00189.safetensors", + "model.layers.94.input_layernorm.weight": "model-00189-of-00189.safetensors", + "model.layers.94.mlp.down_proj.weight": "model-00189-of-00189.safetensors", + "model.layers.94.mlp.down_proj.weight_scale": "model-00189-of-00189.safetensors", + "model.layers.94.mlp.up_proj.weight": "model-00189-of-00189.safetensors", + "model.layers.94.mlp.up_proj.weight_scale": "model-00189-of-00189.safetensors", + "model.layers.94.post_attention_layernorm.bias": "model-00189-of-00189.safetensors", + "model.layers.94.post_attention_layernorm.weight": "model-00189-of-00189.safetensors", + "model.layers.94.self_attn.k_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.94.self_attn.k_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.94.self_attn.o_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.94.self_attn.o_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.94.self_attn.q_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.94.self_attn.q_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.94.self_attn.v_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.94.self_attn.v_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.95.input_layernorm.bias": "model-00189-of-00189.safetensors", + "model.layers.95.input_layernorm.weight": "model-00189-of-00189.safetensors", + "model.layers.95.mlp.down_proj.weight": "model-00189-of-00189.safetensors", + "model.layers.95.mlp.down_proj.weight_scale": "model-00189-of-00189.safetensors", + "model.layers.95.mlp.up_proj.weight": "model-00189-of-00189.safetensors", + "model.layers.95.mlp.up_proj.weight_scale": "model-00189-of-00189.safetensors", + "model.layers.95.post_attention_layernorm.bias": "model-00189-of-00189.safetensors", + "model.layers.95.post_attention_layernorm.weight": "model-00189-of-00189.safetensors", + "model.layers.95.self_attn.k_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.95.self_attn.k_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.95.self_attn.o_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.95.self_attn.o_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.95.self_attn.q_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.95.self_attn.q_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.layers.95.self_attn.v_proj.weight": "model-00187-of-00189.safetensors", + "model.layers.95.self_attn.v_proj.weight_scale": "model-00187-of-00189.safetensors", + "model.norm.bias": "model-00189-of-00189.safetensors", + "model.norm.weight": "model-00189-of-00189.safetensors" + } +} \ No newline at end of file diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000000000000000000000000000000000000..451134b2ddc2e78555d1e857518c54b4bdc2e87d --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "unk_token": { + "content": "", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000000000000000000000000000000000000..a093afc1a84460c37d9af55bb7e8a056851d6c29 --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf6ac4271cd566ddd5cb6e1675a2457bacce5e683b30b35f9668377b36ca2628 +size 21919422 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000000000000000000000000000000000000..f59f15ef480afcbcb6f981ebdda6e95a1eaf7e85 --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,6 @@ +{ + "bos_token": "", + "eos_token": "", + "model_max_length": 4096, + "tokenizer_class": "PreTrainedTokenizerFast" +} \ No newline at end of file