Upload 8 files
Browse files- config.json +32 -0
- generation_config.json +6 -0
- merges.txt +0 -0
- pytorch_model.bin +3 -0
- pytorch_model.safetensors +3 -0
- stripper-gpt2.py +32 -0
- tokenizer.json +0 -0
- vocab.json +0 -0
config.json
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"activation_function": "gelu_new",
|
3 |
+
"architectures": [
|
4 |
+
"GPT2LMHeadModel"
|
5 |
+
],
|
6 |
+
"attn_pdrop": 0.1,
|
7 |
+
"bos_token_id": 50256,
|
8 |
+
"embd_pdrop": 0.1,
|
9 |
+
"eos_token_id": 50256,
|
10 |
+
"initializer_range": 0.02,
|
11 |
+
"layer_norm_epsilon": 1e-05,
|
12 |
+
"model_type": "gpt2",
|
13 |
+
"n_ctx": 1024,
|
14 |
+
"n_embd": 1600,
|
15 |
+
"n_head": 12,
|
16 |
+
"n_layer": 12,
|
17 |
+
"n_positions": 1024,
|
18 |
+
"output_past": true,
|
19 |
+
"resid_pdrop": 0.1,
|
20 |
+
"summary_activation": null,
|
21 |
+
"summary_first_dropout": 0.1,
|
22 |
+
"summary_proj_to_labels": true,
|
23 |
+
"summary_type": "cls_index",
|
24 |
+
"summary_use_proj": true,
|
25 |
+
"task_specific_params": {
|
26 |
+
"text-generation": {
|
27 |
+
"do_sample": true,
|
28 |
+
"max_length": 50
|
29 |
+
}
|
30 |
+
},
|
31 |
+
"vocab_size": 50257
|
32 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"bos_token_id": 50256,
|
4 |
+
"eos_token_id": 50256,
|
5 |
+
"transformers_version": "4.27.0.dev0"
|
6 |
+
}
|
merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:53761927cfd2227ce36499107b66b3dcc2764101bd288d5d46040b84407fcae2
|
3 |
+
size 1854151118
|
pytorch_model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ba108df3f437e29200ba093d2aedd9e877d3d55ddf57b24eb0519e47a753f1a3
|
3 |
+
size 1854115872
|
stripper-gpt2.py
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
import sys
|
3 |
+
from safetensors.torch import save_file
|
4 |
+
|
5 |
+
|
6 |
+
model = torch.load('pytorch_model.bin', map_location="cpu")
|
7 |
+
for key, value in model.items():
|
8 |
+
print(key)
|
9 |
+
# sys.exit() # Comment this out after you've found the key you wish to stop at, and also the final key you wish to keep.
|
10 |
+
|
11 |
+
|
12 |
+
stop_key = 'h.12.ln_1.weight' # The key you want to stop at (the previous key is kept)
|
13 |
+
final_keys = ['ln_f.weight', 'ln_f.bias'] # The final key/keys in the model which get saved.
|
14 |
+
|
15 |
+
|
16 |
+
stripped = {}
|
17 |
+
stopped = False
|
18 |
+
for key, value in model.items():
|
19 |
+
if key == stop_key:
|
20 |
+
stopped = True
|
21 |
+
continue
|
22 |
+
|
23 |
+
if key in final_keys:
|
24 |
+
stripped[key] = value
|
25 |
+
|
26 |
+
if stopped is False:
|
27 |
+
stripped[key] = value
|
28 |
+
|
29 |
+
|
30 |
+
save_file(stripped, 'pytorch_model_stripped.safetensors')
|
31 |
+
torch.save(stripped, 'pytorch_model_stripped.bin')
|
32 |
+
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|