Training in progress, step 500
Browse files- .gitignore +1 -0
- config.json +37 -0
- pytorch_model.bin +3 -0
- runs/Feb03_16-26-49_2cb2cefa7ce2/1675441634.171871/events.out.tfevents.1675441634.2cb2cefa7ce2.9378.1 +3 -0
- runs/Feb03_16-26-49_2cb2cefa7ce2/events.out.tfevents.1675441634.2cb2cefa7ce2.9378.0 +3 -0
- runs/Feb03_17-09-29_2cb2cefa7ce2/1675444176.689174/events.out.tfevents.1675444176.2cb2cefa7ce2.1215.1 +3 -0
- runs/Feb03_17-09-29_2cb2cefa7ce2/events.out.tfevents.1675444176.2cb2cefa7ce2.1215.0 +3 -0
- training_args.bin +3 -0
.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 sha256:0c4db235574bbb7495be7b07d6844c71bbce3914f24f40cac7e0359262b0023f
|
| 3 |
+
size 4600336581
|
runs/Feb03_16-26-49_2cb2cefa7ce2/1675441634.171871/events.out.tfevents.1675441634.2cb2cefa7ce2.9378.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:93bd02fe2563b4e3f99454ffb276dcd61c40f581c96bc046d9e9c94df82f5b9b
|
| 3 |
+
size 5757
|
runs/Feb03_16-26-49_2cb2cefa7ce2/events.out.tfevents.1675441634.2cb2cefa7ce2.9378.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:140a2a692db1a4f41da5bfcadf5ece93cdc8af669b5ad5f28bc3bcbea76809a6
|
| 3 |
+
size 10503
|
runs/Feb03_17-09-29_2cb2cefa7ce2/1675444176.689174/events.out.tfevents.1675444176.2cb2cefa7ce2.1215.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5667c471f536a35594f3bd4ddbd5f3f7de42b68f611da32ba3f034b7ffbe0ed6
|
| 3 |
+
size 5757
|
runs/Feb03_17-09-29_2cb2cefa7ce2/events.out.tfevents.1675444176.2cb2cefa7ce2.1215.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fcc43d91308877774be01af97219fc63e524c21ac03663a0db5c0d1b39c6098d
|
| 3 |
+
size 12344
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3cfbc155b81bc28b76b544e27d417fd99ad289f5a77f11f64f7e1798458a43e
|
| 3 |
+
size 3515
|