Add pytorch model at 320000 steps
Browse files- config.json +2 -2
- flax_to_pytorch.py +1 -1
- pytorch_model.bin +1 -1
- run_gpt.sh +2 -1
- runs/Dec24/events.out.tfevents.1640112885.t1v-n-f9cfcc28-w-0.187197.0.v2 +0 -1
- runs/events.out.tfevents.1640112885.t1v-n-f9cfcc28-w-0.187197.0.v2 +0 -3
- runs/events.out.tfevents.1640332964.t1v-n-f9cfcc28-w-0.384322.0.v2 +2 -2
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "
|
3 |
"activation_function": "gelu_new",
|
4 |
"architectures": [
|
5 |
"GPT2LMHeadModel"
|
@@ -11,7 +11,6 @@
|
|
11 |
"initializer_range": 0.02,
|
12 |
"layer_norm_epsilon": 1e-05,
|
13 |
"model_type": "gpt2",
|
14 |
-
"torch_dtype": "float32",
|
15 |
"n_ctx": 1024,
|
16 |
"n_embd": 1024,
|
17 |
"n_head": 16,
|
@@ -35,6 +34,7 @@
|
|
35 |
"max_length": 50
|
36 |
}
|
37 |
},
|
|
|
38 |
"transformers_version": "4.13.0",
|
39 |
"use_cache": true,
|
40 |
"vocab_size": 50257
|
|
|
1 |
{
|
2 |
+
"_name_or_path": ".",
|
3 |
"activation_function": "gelu_new",
|
4 |
"architectures": [
|
5 |
"GPT2LMHeadModel"
|
|
|
11 |
"initializer_range": 0.02,
|
12 |
"layer_norm_epsilon": 1e-05,
|
13 |
"model_type": "gpt2",
|
|
|
14 |
"n_ctx": 1024,
|
15 |
"n_embd": 1024,
|
16 |
"n_head": 16,
|
|
|
34 |
"max_length": 50
|
35 |
}
|
36 |
},
|
37 |
+
"torch_dtype": "float32",
|
38 |
"transformers_version": "4.13.0",
|
39 |
"use_cache": true,
|
40 |
"vocab_size": 50257
|
flax_to_pytorch.py
CHANGED
@@ -13,7 +13,7 @@ model_fx = FlaxGPT2LMHeadModel.from_pretrained(".")
|
|
13 |
# model_fx.params = to_f32(model_fx.params)
|
14 |
# model_fx.save_pretrained("./fx")
|
15 |
model_pt = GPT2LMHeadModel.from_pretrained(".", from_flax=True)
|
16 |
-
model_pt.save_pretrained("
|
17 |
input_ids = np.asarray(2 * [128 * [0]], dtype=np.int32)
|
18 |
input_ids_pt = torch.tensor(input_ids)
|
19 |
logits_pt = model_pt(input_ids_pt).logits
|
|
|
13 |
# model_fx.params = to_f32(model_fx.params)
|
14 |
# model_fx.save_pretrained("./fx")
|
15 |
model_pt = GPT2LMHeadModel.from_pretrained(".", from_flax=True)
|
16 |
+
model_pt.save_pretrained(".")
|
17 |
input_ids = np.asarray(2 * [128 * [0]], dtype=np.int32)
|
18 |
input_ids_pt = torch.tensor(input_ids)
|
19 |
logits_pt = model_pt(input_ids_pt).logits
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1444576537
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2abd6b504ff54d4765bd621491f332f9f75b1f869c414e5f301d8ec9c054a929
|
3 |
size 1444576537
|
run_gpt.sh
CHANGED
@@ -15,6 +15,7 @@ python run_clm_flax.py \
|
|
15 |
--output_dir="${MODEL_PATH}" \
|
16 |
--model_type="gpt2" \
|
17 |
--config_name="${MODEL_PATH}" \
|
|
|
18 |
--tokenizer_name="${MODEL_PATH}" \
|
19 |
--preprocessing_num_workers="96" \
|
20 |
--do_train --do_eval \
|
@@ -28,7 +29,7 @@ python run_clm_flax.py \
|
|
28 |
--overwrite_output_dir \
|
29 |
--num_train_epochs="1" \
|
30 |
--logging_steps="500" \
|
31 |
-
--save_steps="
|
32 |
--eval_steps="2500"
|
33 |
|
34 |
# \
|
|
|
15 |
--output_dir="${MODEL_PATH}" \
|
16 |
--model_type="gpt2" \
|
17 |
--config_name="${MODEL_PATH}" \
|
18 |
+
--model_name_or_path="${MODEL_PATH}" \
|
19 |
--tokenizer_name="${MODEL_PATH}" \
|
20 |
--preprocessing_num_workers="96" \
|
21 |
--do_train --do_eval \
|
|
|
29 |
--overwrite_output_dir \
|
30 |
--num_train_epochs="1" \
|
31 |
--logging_steps="500" \
|
32 |
+
--save_steps="20001" \
|
33 |
--eval_steps="2500"
|
34 |
|
35 |
# \
|
runs/Dec24/events.out.tfevents.1640112885.t1v-n-f9cfcc28-w-0.187197.0.v2
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
../events.out.tfevents.1640112885.t1v-n-f9cfcc28-w-0.187197.0.v2
|
|
|
|
runs/events.out.tfevents.1640112885.t1v-n-f9cfcc28-w-0.187197.0.v2
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:3f4b57034a982516d8b60834fa3bd1e8af2277469ad9973b3ec86241cb292a89
|
3 |
-
size 17870829
|
|
|
|
|
|
|
|
runs/events.out.tfevents.1640332964.t1v-n-f9cfcc28-w-0.384322.0.v2
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:eda007c25947ad3f969a3fc25f74a6182e0cbfb3e0aadc6ec67019ce38e1e545
|
3 |
+
size 50470447
|