simonosgoode
commited on
Commit
·
3fc6f94
1
Parent(s):
efb8542
End of training
Browse files- .gitignore +1 -0
- config.json +32 -0
- pytorch_model.bin +3 -0
- runs/Oct13_13-18-42_129-146-54-156/1665667130.8790777/events.out.tfevents.1665667130.129-146-54-156.64725.1 +3 -0
- runs/Oct13_13-18-42_129-146-54-156/events.out.tfevents.1665667130.129-146-54-156.64725.0 +3 -0
- runs/Oct13_13-18-42_129-146-54-156/events.out.tfevents.1665678609.129-146-54-156.64725.2 +3 -0
- training_args.bin +3 -0
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
checkpoint-*/
|
config.json
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "simonosgoode/bloom-560m-finetuned-cdn_law",
|
3 |
+
"apply_residual_connection_post_layernorm": false,
|
4 |
+
"architectures": [
|
5 |
+
"BloomForCausalLM"
|
6 |
+
],
|
7 |
+
"attention_dropout": 0.0,
|
8 |
+
"attention_softmax_in_fp32": true,
|
9 |
+
"bias_dropout_fusion": true,
|
10 |
+
"bos_token_id": 1,
|
11 |
+
"eos_token_id": 2,
|
12 |
+
"hidden_dropout": 0.0,
|
13 |
+
"hidden_size": 1024,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"layer_norm_epsilon": 1e-05,
|
16 |
+
"masked_softmax_fusion": true,
|
17 |
+
"model_type": "bloom",
|
18 |
+
"n_head": 16,
|
19 |
+
"n_inner": null,
|
20 |
+
"n_layer": 24,
|
21 |
+
"offset_alibi": 100,
|
22 |
+
"pad_token_id": 3,
|
23 |
+
"pretraining_tp": 1,
|
24 |
+
"skip_bias_add": true,
|
25 |
+
"skip_bias_add_qkv": false,
|
26 |
+
"slow_but_exact": false,
|
27 |
+
"torch_dtype": "float32",
|
28 |
+
"transformers_version": "4.23.1",
|
29 |
+
"unk_token_id": 0,
|
30 |
+
"use_cache": true,
|
31 |
+
"vocab_size": 250880
|
32 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b556607a90ab80b666d5d020ab36a0c380080e7a9ac28e992bd95b14e6a8aee0
|
3 |
+
size 2236955191
|
runs/Oct13_13-18-42_129-146-54-156/1665667130.8790777/events.out.tfevents.1665667130.129-146-54-156.64725.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d83527e1321be07deb004f1c9e608a0d0d01bf51e623ea1515cf44b269825e8b
|
3 |
+
size 5578
|
runs/Oct13_13-18-42_129-146-54-156/events.out.tfevents.1665667130.129-146-54-156.64725.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:83da32f868d3bf6b578202ea42ded01c624a984aa9cd98db637b8eea0babc427
|
3 |
+
size 5639
|
runs/Oct13_13-18-42_129-146-54-156/events.out.tfevents.1665678609.129-146-54-156.64725.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1821ac49ce1b739fb275923ccae25f8be60c1a26693e8b19ba57fe77c1994e38
|
3 |
+
size 316
|
training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6c26f401e39c703f3a0641beb61c3c1362f2765b8fc6a0892b72a374243cab84
|
3 |
+
size 3439
|