Training in progress, step 500
Browse files
config.json
ADDED
|
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_function": "gelu_new",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"GPTNeoForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0,
|
| 7 |
+
"attention_layers": [
|
| 8 |
+
"global",
|
| 9 |
+
"local",
|
| 10 |
+
"global",
|
| 11 |
+
"local",
|
| 12 |
+
"global",
|
| 13 |
+
"local",
|
| 14 |
+
"global",
|
| 15 |
+
"local",
|
| 16 |
+
"global",
|
| 17 |
+
"local",
|
| 18 |
+
"global",
|
| 19 |
+
"local"
|
| 20 |
+
],
|
| 21 |
+
"attention_types": [
|
| 22 |
+
[
|
| 23 |
+
[
|
| 24 |
+
"global",
|
| 25 |
+
"local"
|
| 26 |
+
],
|
| 27 |
+
6
|
| 28 |
+
]
|
| 29 |
+
],
|
| 30 |
+
"bos_token_id": 50256,
|
| 31 |
+
"classifier_dropout": 0.1,
|
| 32 |
+
"dtype": "float32",
|
| 33 |
+
"embed_dropout": 0,
|
| 34 |
+
"eos_token_id": 50256,
|
| 35 |
+
"gradient_checkpointing": false,
|
| 36 |
+
"hidden_size": 768,
|
| 37 |
+
"initializer_range": 0.02,
|
| 38 |
+
"intermediate_size": null,
|
| 39 |
+
"layer_norm_epsilon": 1e-05,
|
| 40 |
+
"max_position_embeddings": 2048,
|
| 41 |
+
"model_type": "gpt_neo",
|
| 42 |
+
"num_heads": 12,
|
| 43 |
+
"num_layers": 12,
|
| 44 |
+
"resid_dropout": 0,
|
| 45 |
+
"summary_activation": null,
|
| 46 |
+
"summary_first_dropout": 0.1,
|
| 47 |
+
"summary_proj_to_labels": true,
|
| 48 |
+
"summary_type": "cls_index",
|
| 49 |
+
"summary_use_proj": true,
|
| 50 |
+
"transformers_version": "4.56.1",
|
| 51 |
+
"use_cache": true,
|
| 52 |
+
"vocab_size": 50257,
|
| 53 |
+
"window_size": 256
|
| 54 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:166db36084a8b94b8c1348bb7a1d413dfdbf6132367af6eeef68ade457367599
|
| 3 |
+
size 500811336
|
runs/Sep29_16-26-54_4c8d3689d075/events.out.tfevents.1759163221.4c8d3689d075.3006.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:749cc61eab72c2a196b6146356957fa135c8c30aced96fed1b1462e0b5700433
|
| 3 |
+
size 5644
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4be17abadc52c5879775f04c28cc84e66d6995ba8f1b0991f40f1c193863d30b
|
| 3 |
+
size 5841
|