Training in progress, step 500
Browse files
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
checkpoint-*/
|
config.json
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "bigcode/santacoder",
|
3 |
+
"activation_function": "gelu_fast",
|
4 |
+
"architectures": [
|
5 |
+
"GPT2LMHeadCustomModel"
|
6 |
+
],
|
7 |
+
"attention_head_type": "multiquery",
|
8 |
+
"attn_pdrop": 0.1,
|
9 |
+
"auto_map": {
|
10 |
+
"AutoConfig": "configuration_gpt2_mq.GPT2CustomConfig",
|
11 |
+
"AutoModelForCausalLM": "modeling_gpt2_mq.GPT2LMHeadCustomModel"
|
12 |
+
},
|
13 |
+
"bos_token_id": 50256,
|
14 |
+
"embd_pdrop": 0.1,
|
15 |
+
"eos_token_id": 50256,
|
16 |
+
"initializer_range": 0.02,
|
17 |
+
"layer_norm_epsilon": 1e-05,
|
18 |
+
"model_type": "gpt2",
|
19 |
+
"n_embd": 2048,
|
20 |
+
"n_head": 16,
|
21 |
+
"n_inner": 8192,
|
22 |
+
"n_layer": 24,
|
23 |
+
"n_positions": 2048,
|
24 |
+
"reorder_and_upcast_attn": false,
|
25 |
+
"resid_pdrop": 0.1,
|
26 |
+
"scale_attn_by_inverse_layer_idx": false,
|
27 |
+
"scale_attn_weights": true,
|
28 |
+
"summary_activation": null,
|
29 |
+
"summary_first_dropout": 0.1,
|
30 |
+
"summary_proj_to_labels": true,
|
31 |
+
"summary_type": "cls_index",
|
32 |
+
"summary_use_proj": true,
|
33 |
+
"torch_dtype": "float32",
|
34 |
+
"transformers_version": "4.26.0.dev0",
|
35 |
+
"use_cache": false,
|
36 |
+
"vocab_size": 49280
|
37 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5e552aef00c6995303b3dfe39b2418cc27056bd1c9dfabb7620cf40ecfdeb4ec
|
3 |
+
size 4600336581
|
runs/Jan26_15-38-39_2cb2cefa7ce2/1674747529.794282/events.out.tfevents.1674747529.2cb2cefa7ce2.21439.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:93a9a94263b782f0726883e62384cc4916601fe7dd8403fe5d43edacfafcc3fd
|
3 |
+
size 5754
|
runs/Jan26_15-38-39_2cb2cefa7ce2/events.out.tfevents.1674747529.2cb2cefa7ce2.21439.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:db0188d6f52c0d6d3e796d0fe5ba9f640251b76be0a2e9814142d652d010e705
|
3 |
+
size 12341
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9029018ffe265d96b927559442e73b23140125b10887d248e12625540ed38f1d
|
3 |
+
size 3515
|