CrashSAM
commited on
Commit
•
2a3162e
1
Parent(s):
fd15a99
Upload config
Browse files- config.json +74 -0
config.json
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"act_fn": "gelu",
|
| 3 |
+
"alibi": true,
|
| 4 |
+
"alibi_bias_max": 8,
|
| 5 |
+
"architectures": [
|
| 6 |
+
"MPTForCausalLM"
|
| 7 |
+
],
|
| 8 |
+
"attn_config": {
|
| 9 |
+
"alibi": true,
|
| 10 |
+
"alibi_bias_max": 8,
|
| 11 |
+
"attn_impl": "torch",
|
| 12 |
+
"attn_pdrop": 0,
|
| 13 |
+
"attn_type": "multihead_attention",
|
| 14 |
+
"attn_uses_sequence_id": false,
|
| 15 |
+
"clip_qkv": null,
|
| 16 |
+
"prefix_lm": false,
|
| 17 |
+
"qk_ln": false,
|
| 18 |
+
"softmax_scale": null
|
| 19 |
+
},
|
| 20 |
+
"attn_impl": "torch",
|
| 21 |
+
"attn_pdrop": 0,
|
| 22 |
+
"attn_type": "multihead_attention",
|
| 23 |
+
"attn_uses_sequence_id": false,
|
| 24 |
+
"auto_map": {
|
| 25 |
+
"AutoConfig": "mosaicml/mpt-7b--configuration_mpt.MPTConfig",
|
| 26 |
+
"AutoModelForCausalLM": "mosaicml/mpt-7b--modeling_mpt.MPTForCausalLM"
|
| 27 |
+
},
|
| 28 |
+
"clip_qkv": null,
|
| 29 |
+
"d_model": 4096,
|
| 30 |
+
"emb_pdrop": 0,
|
| 31 |
+
"emb_prob_drop": 0.0,
|
| 32 |
+
"embedding_fraction": 1.0,
|
| 33 |
+
"expansion_ratio": 4,
|
| 34 |
+
"flash_attn_key_chunk_size": 2048,
|
| 35 |
+
"flash_attn_query_chunk_size": 1024,
|
| 36 |
+
"from_pt": false,
|
| 37 |
+
"gradient_checkpointing": "nothing_saveable",
|
| 38 |
+
"init_config": {
|
| 39 |
+
"emb_init_std": null,
|
| 40 |
+
"emb_init_uniform_lim": null,
|
| 41 |
+
"fan_mode": "fan_in",
|
| 42 |
+
"init_div_is_residual": true,
|
| 43 |
+
"init_gain": 0,
|
| 44 |
+
"init_nonlinearity": "relu",
|
| 45 |
+
"init_std": 0.02,
|
| 46 |
+
"name": "kaiming_normal_",
|
| 47 |
+
"verbose": 0
|
| 48 |
+
},
|
| 49 |
+
"init_device": "cpu",
|
| 50 |
+
"learned_pos_emb": true,
|
| 51 |
+
"logit_scale": null,
|
| 52 |
+
"max_seq_len": 2048,
|
| 53 |
+
"model_type": "mpt",
|
| 54 |
+
"n_heads": 32,
|
| 55 |
+
"n_layers": 32,
|
| 56 |
+
"no_bias": true,
|
| 57 |
+
"norm_type": "low_precision_layernorm",
|
| 58 |
+
"prefix_lm": false,
|
| 59 |
+
"qk_ln": false,
|
| 60 |
+
"resid_pdrop": 0,
|
| 61 |
+
"resid_prob_drop": 0.0,
|
| 62 |
+
"softmax_scale": null,
|
| 63 |
+
"tokenizer_name": "EleutherAI/gpt-neox-20b",
|
| 64 |
+
"torch_dtype": "bfloat16",
|
| 65 |
+
"transformers_version": "4.31.0",
|
| 66 |
+
"use_bias": false,
|
| 67 |
+
"use_cache": false,
|
| 68 |
+
"use_flash_attention": false,
|
| 69 |
+
"use_lm_head": false,
|
| 70 |
+
"use_norm_bias": false,
|
| 71 |
+
"use_pjit_attention_force": false,
|
| 72 |
+
"verbose": 0,
|
| 73 |
+
"vocab_size": 50432
|
| 74 |
+
}
|