Commit
•
f1355d8
1
Parent(s):
d328d4d
Merge with linear
Browse files- config.json +5 -2
- pytorch_model.bin +1 -1
config.json
CHANGED
|
@@ -3,11 +3,13 @@
|
|
| 3 |
"architectures": [
|
| 4 |
"GPTBigCodeLMHeadModel"
|
| 5 |
],
|
|
|
|
| 6 |
"attention_type": 2,
|
| 7 |
"attn_pdrop": 0.1,
|
| 8 |
"bos_token_id": 50256,
|
| 9 |
"embd_pdrop": 0.1,
|
| 10 |
"eos_token_id": 50256,
|
|
|
|
| 11 |
"initializer_range": 0.02,
|
| 12 |
"layer_norm_epsilon": 1e-05,
|
| 13 |
"model_type": "gpt_bigcode",
|
|
@@ -16,9 +18,9 @@
|
|
| 16 |
"n_inner": 8192,
|
| 17 |
"n_layer": 24,
|
| 18 |
"n_positions": 2048,
|
| 19 |
-
"reorder_and_upcast_attn": false,
|
| 20 |
"resid_pdrop": 0.1,
|
| 21 |
-
"
|
|
|
|
| 22 |
"scale_attn_weights": true,
|
| 23 |
"summary_activation": null,
|
| 24 |
"summary_first_dropout": 0.1,
|
|
@@ -27,5 +29,6 @@
|
|
| 27 |
"summary_use_proj": true,
|
| 28 |
"transformers_version": "4.27.0.dev0",
|
| 29 |
"use_cache": true,
|
|
|
|
| 30 |
"vocab_size": 49280
|
| 31 |
}
|
|
|
|
| 3 |
"architectures": [
|
| 4 |
"GPTBigCodeLMHeadModel"
|
| 5 |
],
|
| 6 |
+
"attention_softmax_in_fp32": true,
|
| 7 |
"attention_type": 2,
|
| 8 |
"attn_pdrop": 0.1,
|
| 9 |
"bos_token_id": 50256,
|
| 10 |
"embd_pdrop": 0.1,
|
| 11 |
"eos_token_id": 50256,
|
| 12 |
+
"inference_runner": 0,
|
| 13 |
"initializer_range": 0.02,
|
| 14 |
"layer_norm_epsilon": 1e-05,
|
| 15 |
"model_type": "gpt_bigcode",
|
|
|
|
| 18 |
"n_inner": 8192,
|
| 19 |
"n_layer": 24,
|
| 20 |
"n_positions": 2048,
|
|
|
|
| 21 |
"resid_pdrop": 0.1,
|
| 22 |
+
"runner_max_sequence_length": null,
|
| 23 |
+
"scale_attention_softmax_in_fp32": true,
|
| 24 |
"scale_attn_weights": true,
|
| 25 |
"summary_activation": null,
|
| 26 |
"summary_first_dropout": 0.1,
|
|
|
|
| 29 |
"summary_use_proj": true,
|
| 30 |
"transformers_version": "4.27.0.dev0",
|
| 31 |
"use_cache": true,
|
| 32 |
+
"validate_runner_input": true,
|
| 33 |
"vocab_size": 49280
|
| 34 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2249864093
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:50d0abae8bfd19082f125e7a40567cba71de296637802eb70157fe55fa66923f
|
| 3 |
size 2249864093
|