Upload folder using huggingface_hub
Browse files- config.json +46 -0
- generation_config.json +6 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +1085 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "distilbert/distilgpt2",
|
| 3 |
+
"_num_labels": 1,
|
| 4 |
+
"activation_function": "gelu_new",
|
| 5 |
+
"architectures": [
|
| 6 |
+
"GPT2LMHeadModel"
|
| 7 |
+
],
|
| 8 |
+
"attn_pdrop": 0.1,
|
| 9 |
+
"bos_token_id": 50256,
|
| 10 |
+
"embd_pdrop": 0.1,
|
| 11 |
+
"eos_token_id": 50256,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "LABEL_0"
|
| 14 |
+
},
|
| 15 |
+
"initializer_range": 0.02,
|
| 16 |
+
"label2id": {
|
| 17 |
+
"LABEL_0": 0
|
| 18 |
+
},
|
| 19 |
+
"layer_norm_epsilon": 1e-05,
|
| 20 |
+
"model_type": "gpt2",
|
| 21 |
+
"n_ctx": 1024,
|
| 22 |
+
"n_embd": 768,
|
| 23 |
+
"n_head": 12,
|
| 24 |
+
"n_inner": null,
|
| 25 |
+
"n_layer": 6,
|
| 26 |
+
"n_positions": 1024,
|
| 27 |
+
"reorder_and_upcast_attn": false,
|
| 28 |
+
"resid_pdrop": 0.1,
|
| 29 |
+
"scale_attn_by_inverse_layer_idx": false,
|
| 30 |
+
"scale_attn_weights": true,
|
| 31 |
+
"summary_activation": null,
|
| 32 |
+
"summary_first_dropout": 0.1,
|
| 33 |
+
"summary_proj_to_labels": true,
|
| 34 |
+
"summary_type": "cls_index",
|
| 35 |
+
"summary_use_proj": true,
|
| 36 |
+
"task_specific_params": {
|
| 37 |
+
"text-generation": {
|
| 38 |
+
"do_sample": true,
|
| 39 |
+
"max_length": 50
|
| 40 |
+
}
|
| 41 |
+
},
|
| 42 |
+
"torch_dtype": "float32",
|
| 43 |
+
"transformers_version": "4.38.2",
|
| 44 |
+
"use_cache": true,
|
| 45 |
+
"vocab_size": 50257
|
| 46 |
+
}
|
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": "4.38.2"
|
| 6 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be8c4250cf32042073bdd8398499d0c077dd9f1609d1c56f50ea839574eda95b
|
| 3 |
+
size 327657928
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7a86878426b0cde4cc43067288fbb7c81062595c51e2fb9349c1171c68d8b552
|
| 3 |
+
size 655364474
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c9c546d1455d9a580cd5d64d5fda722bd985c483d6e0f11e4132a6eaf9135c2a
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b0913d2629e0e5ec4d9824083abef171658925a110f73e93f07ee323cfe6363
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1085 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 3.8748562335968018,
|
| 3 |
+
"best_model_checkpoint": "/kaggle/working/checkpoint-669",
|
| 4 |
+
"epoch": 50.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 11150,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_loss": 3.875725507736206,
|
| 14 |
+
"eval_mean_perplexity": 366.3703887939453,
|
| 15 |
+
"eval_perplexities": [
|
| 16 |
+
335.80645751953125,
|
| 17 |
+
377.47796630859375,
|
| 18 |
+
475.9695129394531,
|
| 19 |
+
276.7535705566406,
|
| 20 |
+
294.2838134765625,
|
| 21 |
+
264.73394775390625,
|
| 22 |
+
394.9538269042969,
|
| 23 |
+
325.2862854003906,
|
| 24 |
+
525.1426391601562,
|
| 25 |
+
393.2958679199219
|
| 26 |
+
],
|
| 27 |
+
"eval_runtime": 2.3638,
|
| 28 |
+
"eval_samples_per_second": 4.231,
|
| 29 |
+
"eval_steps_per_second": 0.846,
|
| 30 |
+
"step": 223
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 2.0,
|
| 34 |
+
"eval_loss": 3.877554416656494,
|
| 35 |
+
"eval_mean_perplexity": 373.7104461669922,
|
| 36 |
+
"eval_perplexities": [
|
| 37 |
+
356.83929443359375,
|
| 38 |
+
425.78094482421875,
|
| 39 |
+
495.7661437988281,
|
| 40 |
+
287.1143798828125,
|
| 41 |
+
270.86737060546875,
|
| 42 |
+
256.76934814453125,
|
| 43 |
+
391.27044677734375,
|
| 44 |
+
345.6097412109375,
|
| 45 |
+
498.4945373535156,
|
| 46 |
+
408.5922546386719
|
| 47 |
+
],
|
| 48 |
+
"eval_runtime": 2.2583,
|
| 49 |
+
"eval_samples_per_second": 4.428,
|
| 50 |
+
"eval_steps_per_second": 0.886,
|
| 51 |
+
"step": 446
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 3.0,
|
| 55 |
+
"eval_loss": 3.8748562335968018,
|
| 56 |
+
"eval_mean_perplexity": 378.68709869384764,
|
| 57 |
+
"eval_perplexities": [
|
| 58 |
+
404.7926940917969,
|
| 59 |
+
400.53558349609375,
|
| 60 |
+
487.69427490234375,
|
| 61 |
+
229.20298767089844,
|
| 62 |
+
278.294189453125,
|
| 63 |
+
259.6055908203125,
|
| 64 |
+
387.93963623046875,
|
| 65 |
+
365.1226501464844,
|
| 66 |
+
508.79669189453125,
|
| 67 |
+
464.8866882324219
|
| 68 |
+
],
|
| 69 |
+
"eval_runtime": 2.4538,
|
| 70 |
+
"eval_samples_per_second": 4.075,
|
| 71 |
+
"eval_steps_per_second": 0.815,
|
| 72 |
+
"step": 669
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 4.0,
|
| 76 |
+
"eval_loss": 3.889125108718872,
|
| 77 |
+
"eval_mean_perplexity": 395.6389587402344,
|
| 78 |
+
"eval_perplexities": [
|
| 79 |
+
386.92559814453125,
|
| 80 |
+
444.32073974609375,
|
| 81 |
+
480.409912109375,
|
| 82 |
+
278.6746520996094,
|
| 83 |
+
279.8287353515625,
|
| 84 |
+
286.8155212402344,
|
| 85 |
+
426.2447204589844,
|
| 86 |
+
352.7798767089844,
|
| 87 |
+
545.115234375,
|
| 88 |
+
475.27459716796875
|
| 89 |
+
],
|
| 90 |
+
"eval_runtime": 2.2494,
|
| 91 |
+
"eval_samples_per_second": 4.446,
|
| 92 |
+
"eval_steps_per_second": 0.889,
|
| 93 |
+
"step": 892
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 5.0,
|
| 97 |
+
"eval_loss": 3.9018893241882324,
|
| 98 |
+
"eval_mean_perplexity": 418.4004302978516,
|
| 99 |
+
"eval_perplexities": [
|
| 100 |
+
426.7615051269531,
|
| 101 |
+
474.9723815917969,
|
| 102 |
+
516.2012329101562,
|
| 103 |
+
285.6946105957031,
|
| 104 |
+
314.9617919921875,
|
| 105 |
+
305.4250793457031,
|
| 106 |
+
411.5309143066406,
|
| 107 |
+
412.8928527832031,
|
| 108 |
+
537.16943359375,
|
| 109 |
+
498.3945007324219
|
| 110 |
+
],
|
| 111 |
+
"eval_runtime": 2.2536,
|
| 112 |
+
"eval_samples_per_second": 4.437,
|
| 113 |
+
"eval_steps_per_second": 0.887,
|
| 114 |
+
"step": 1115
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 6.0,
|
| 118 |
+
"eval_loss": 3.9105305671691895,
|
| 119 |
+
"eval_mean_perplexity": 423.3804229736328,
|
| 120 |
+
"eval_perplexities": [
|
| 121 |
+
421.5846862792969,
|
| 122 |
+
457.750244140625,
|
| 123 |
+
521.7881469726562,
|
| 124 |
+
293.4595642089844,
|
| 125 |
+
283.9613037109375,
|
| 126 |
+
287.46807861328125,
|
| 127 |
+
451.5904846191406,
|
| 128 |
+
458.28509521484375,
|
| 129 |
+
608.7252197265625,
|
| 130 |
+
449.19140625
|
| 131 |
+
],
|
| 132 |
+
"eval_runtime": 2.261,
|
| 133 |
+
"eval_samples_per_second": 4.423,
|
| 134 |
+
"eval_steps_per_second": 0.885,
|
| 135 |
+
"step": 1338
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 7.0,
|
| 139 |
+
"eval_loss": 3.910905122756958,
|
| 140 |
+
"eval_mean_perplexity": 437.6560882568359,
|
| 141 |
+
"eval_perplexities": [
|
| 142 |
+
459.6704406738281,
|
| 143 |
+
471.1756591796875,
|
| 144 |
+
508.8789367675781,
|
| 145 |
+
278.22161865234375,
|
| 146 |
+
305.4996337890625,
|
| 147 |
+
335.0756530761719,
|
| 148 |
+
449.5406494140625,
|
| 149 |
+
402.190673828125,
|
| 150 |
+
618.786376953125,
|
| 151 |
+
547.521240234375
|
| 152 |
+
],
|
| 153 |
+
"eval_runtime": 2.4541,
|
| 154 |
+
"eval_samples_per_second": 4.075,
|
| 155 |
+
"eval_steps_per_second": 0.815,
|
| 156 |
+
"step": 1561
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 8.0,
|
| 160 |
+
"eval_loss": 3.9314892292022705,
|
| 161 |
+
"eval_mean_perplexity": 451.49346618652345,
|
| 162 |
+
"eval_perplexities": [
|
| 163 |
+
432.6178283691406,
|
| 164 |
+
494.5645446777344,
|
| 165 |
+
518.4781494140625,
|
| 166 |
+
289.39727783203125,
|
| 167 |
+
317.63031005859375,
|
| 168 |
+
327.0579528808594,
|
| 169 |
+
491.4974670410156,
|
| 170 |
+
445.9523620605469,
|
| 171 |
+
667.2817993164062,
|
| 172 |
+
530.4569702148438
|
| 173 |
+
],
|
| 174 |
+
"eval_runtime": 3.1442,
|
| 175 |
+
"eval_samples_per_second": 3.18,
|
| 176 |
+
"eval_steps_per_second": 0.636,
|
| 177 |
+
"step": 1784
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 9.0,
|
| 181 |
+
"eval_loss": 3.9446353912353516,
|
| 182 |
+
"eval_mean_perplexity": 451.08067626953124,
|
| 183 |
+
"eval_perplexities": [
|
| 184 |
+
452.6179504394531,
|
| 185 |
+
445.2970275878906,
|
| 186 |
+
576.3162841796875,
|
| 187 |
+
262.3009033203125,
|
| 188 |
+
326.6243591308594,
|
| 189 |
+
311.2583923339844,
|
| 190 |
+
440.389404296875,
|
| 191 |
+
448.5811767578125,
|
| 192 |
+
677.4688110351562,
|
| 193 |
+
569.9524536132812
|
| 194 |
+
],
|
| 195 |
+
"eval_runtime": 2.2555,
|
| 196 |
+
"eval_samples_per_second": 4.434,
|
| 197 |
+
"eval_steps_per_second": 0.887,
|
| 198 |
+
"step": 2007
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 10.0,
|
| 202 |
+
"eval_loss": 3.979356288909912,
|
| 203 |
+
"eval_mean_perplexity": 466.0081390380859,
|
| 204 |
+
"eval_perplexities": [
|
| 205 |
+
479.5843505859375,
|
| 206 |
+
542.103759765625,
|
| 207 |
+
561.8331298828125,
|
| 208 |
+
283.5645751953125,
|
| 209 |
+
342.3445739746094,
|
| 210 |
+
315.0624694824219,
|
| 211 |
+
494.7655334472656,
|
| 212 |
+
397.63812255859375,
|
| 213 |
+
683.822265625,
|
| 214 |
+
559.3626098632812
|
| 215 |
+
],
|
| 216 |
+
"eval_runtime": 2.255,
|
| 217 |
+
"eval_samples_per_second": 4.435,
|
| 218 |
+
"eval_steps_per_second": 0.887,
|
| 219 |
+
"step": 2230
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 11.0,
|
| 223 |
+
"eval_loss": 3.985069751739502,
|
| 224 |
+
"eval_mean_perplexity": 471.23480529785155,
|
| 225 |
+
"eval_perplexities": [
|
| 226 |
+
442.65118408203125,
|
| 227 |
+
578.3666381835938,
|
| 228 |
+
530.2559204101562,
|
| 229 |
+
292.97174072265625,
|
| 230 |
+
325.9568786621094,
|
| 231 |
+
331.6105651855469,
|
| 232 |
+
466.1777648925781,
|
| 233 |
+
453.5052490234375,
|
| 234 |
+
693.37646484375,
|
| 235 |
+
597.4756469726562
|
| 236 |
+
],
|
| 237 |
+
"eval_runtime": 2.2568,
|
| 238 |
+
"eval_samples_per_second": 4.431,
|
| 239 |
+
"eval_steps_per_second": 0.886,
|
| 240 |
+
"step": 2453
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 12.0,
|
| 244 |
+
"eval_loss": 4.011897087097168,
|
| 245 |
+
"eval_mean_perplexity": 481.7092010498047,
|
| 246 |
+
"eval_perplexities": [
|
| 247 |
+
490.259765625,
|
| 248 |
+
590.09716796875,
|
| 249 |
+
524.1170043945312,
|
| 250 |
+
292.7325134277344,
|
| 251 |
+
362.09210205078125,
|
| 252 |
+
320.8348388671875,
|
| 253 |
+
483.432861328125,
|
| 254 |
+
492.00469970703125,
|
| 255 |
+
676.3094482421875,
|
| 256 |
+
585.2116088867188
|
| 257 |
+
],
|
| 258 |
+
"eval_runtime": 2.2714,
|
| 259 |
+
"eval_samples_per_second": 4.402,
|
| 260 |
+
"eval_steps_per_second": 0.88,
|
| 261 |
+
"step": 2676
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 13.0,
|
| 265 |
+
"eval_loss": 4.028537750244141,
|
| 266 |
+
"eval_mean_perplexity": 491.83470153808594,
|
| 267 |
+
"eval_perplexities": [
|
| 268 |
+
434.7430725097656,
|
| 269 |
+
579.150390625,
|
| 270 |
+
531.6643676757812,
|
| 271 |
+
323.02447509765625,
|
| 272 |
+
381.5349426269531,
|
| 273 |
+
329.8095397949219,
|
| 274 |
+
569.4808349609375,
|
| 275 |
+
454.2198486328125,
|
| 276 |
+
723.43798828125,
|
| 277 |
+
591.2815551757812
|
| 278 |
+
],
|
| 279 |
+
"eval_runtime": 2.268,
|
| 280 |
+
"eval_samples_per_second": 4.409,
|
| 281 |
+
"eval_steps_per_second": 0.882,
|
| 282 |
+
"step": 2899
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 14.0,
|
| 286 |
+
"eval_loss": 4.046214580535889,
|
| 287 |
+
"eval_mean_perplexity": 488.7431976318359,
|
| 288 |
+
"eval_perplexities": [
|
| 289 |
+
465.3315734863281,
|
| 290 |
+
577.7365112304688,
|
| 291 |
+
532.9083251953125,
|
| 292 |
+
298.2411193847656,
|
| 293 |
+
371.57415771484375,
|
| 294 |
+
335.2172546386719,
|
| 295 |
+
547.2592163085938,
|
| 296 |
+
480.532470703125,
|
| 297 |
+
732.7112426757812,
|
| 298 |
+
545.9201049804688
|
| 299 |
+
],
|
| 300 |
+
"eval_runtime": 2.2446,
|
| 301 |
+
"eval_samples_per_second": 4.455,
|
| 302 |
+
"eval_steps_per_second": 0.891,
|
| 303 |
+
"step": 3122
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 15.0,
|
| 307 |
+
"eval_loss": 4.064830780029297,
|
| 308 |
+
"eval_mean_perplexity": 493.6821044921875,
|
| 309 |
+
"eval_perplexities": [
|
| 310 |
+
508.1759948730469,
|
| 311 |
+
588.5740356445312,
|
| 312 |
+
494.7876892089844,
|
| 313 |
+
310.3844299316406,
|
| 314 |
+
357.8840026855469,
|
| 315 |
+
349.0634460449219,
|
| 316 |
+
574.1718139648438,
|
| 317 |
+
489.5400085449219,
|
| 318 |
+
730.5029907226562,
|
| 319 |
+
533.7366333007812
|
| 320 |
+
],
|
| 321 |
+
"eval_runtime": 2.2496,
|
| 322 |
+
"eval_samples_per_second": 4.445,
|
| 323 |
+
"eval_steps_per_second": 0.889,
|
| 324 |
+
"step": 3345
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 16.0,
|
| 328 |
+
"eval_loss": 4.0865068435668945,
|
| 329 |
+
"eval_mean_perplexity": 527.1822723388672,
|
| 330 |
+
"eval_perplexities": [
|
| 331 |
+
516.6371459960938,
|
| 332 |
+
603.1266479492188,
|
| 333 |
+
558.2211303710938,
|
| 334 |
+
320.877685546875,
|
| 335 |
+
440.86993408203125,
|
| 336 |
+
384.0972595214844,
|
| 337 |
+
586.449951171875,
|
| 338 |
+
441.53643798828125,
|
| 339 |
+
805.31201171875,
|
| 340 |
+
614.6945190429688
|
| 341 |
+
],
|
| 342 |
+
"eval_runtime": 2.2935,
|
| 343 |
+
"eval_samples_per_second": 4.36,
|
| 344 |
+
"eval_steps_per_second": 0.872,
|
| 345 |
+
"step": 3568
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 17.0,
|
| 349 |
+
"eval_loss": 4.112179756164551,
|
| 350 |
+
"eval_mean_perplexity": 517.5738647460937,
|
| 351 |
+
"eval_perplexities": [
|
| 352 |
+
484.16485595703125,
|
| 353 |
+
623.9596557617188,
|
| 354 |
+
544.3837890625,
|
| 355 |
+
327.7882995605469,
|
| 356 |
+
432.2279052734375,
|
| 357 |
+
325.2957458496094,
|
| 358 |
+
587.2503051757812,
|
| 359 |
+
500.35589599609375,
|
| 360 |
+
768.452392578125,
|
| 361 |
+
581.8598022460938
|
| 362 |
+
],
|
| 363 |
+
"eval_runtime": 2.6951,
|
| 364 |
+
"eval_samples_per_second": 3.71,
|
| 365 |
+
"eval_steps_per_second": 0.742,
|
| 366 |
+
"step": 3791
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 18.0,
|
| 370 |
+
"eval_loss": 4.133630275726318,
|
| 371 |
+
"eval_mean_perplexity": 532.7662231445313,
|
| 372 |
+
"eval_perplexities": [
|
| 373 |
+
459.6676025390625,
|
| 374 |
+
644.7689819335938,
|
| 375 |
+
559.6870727539062,
|
| 376 |
+
313.52886962890625,
|
| 377 |
+
407.5067443847656,
|
| 378 |
+
358.4698486328125,
|
| 379 |
+
635.401123046875,
|
| 380 |
+
511.8723449707031,
|
| 381 |
+
728.2614135742188,
|
| 382 |
+
708.4982299804688
|
| 383 |
+
],
|
| 384 |
+
"eval_runtime": 2.2702,
|
| 385 |
+
"eval_samples_per_second": 4.405,
|
| 386 |
+
"eval_steps_per_second": 0.881,
|
| 387 |
+
"step": 4014
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 19.0,
|
| 391 |
+
"eval_loss": 4.14639949798584,
|
| 392 |
+
"eval_mean_perplexity": 546.8367309570312,
|
| 393 |
+
"eval_perplexities": [
|
| 394 |
+
513.874267578125,
|
| 395 |
+
633.9168090820312,
|
| 396 |
+
623.620849609375,
|
| 397 |
+
317.8645324707031,
|
| 398 |
+
417.3320007324219,
|
| 399 |
+
351.769287109375,
|
| 400 |
+
557.8528442382812,
|
| 401 |
+
530.8079223632812,
|
| 402 |
+
834.164306640625,
|
| 403 |
+
687.1644897460938
|
| 404 |
+
],
|
| 405 |
+
"eval_runtime": 2.28,
|
| 406 |
+
"eval_samples_per_second": 4.386,
|
| 407 |
+
"eval_steps_per_second": 0.877,
|
| 408 |
+
"step": 4237
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 20.0,
|
| 412 |
+
"eval_loss": 4.1683173179626465,
|
| 413 |
+
"eval_mean_perplexity": 530.1545104980469,
|
| 414 |
+
"eval_perplexities": [
|
| 415 |
+
518.4563598632812,
|
| 416 |
+
616.3810424804688,
|
| 417 |
+
584.39404296875,
|
| 418 |
+
357.87530517578125,
|
| 419 |
+
424.8235778808594,
|
| 420 |
+
363.0649719238281,
|
| 421 |
+
572.78369140625,
|
| 422 |
+
489.98748779296875,
|
| 423 |
+
762.8850708007812,
|
| 424 |
+
610.8935546875
|
| 425 |
+
],
|
| 426 |
+
"eval_runtime": 2.5117,
|
| 427 |
+
"eval_samples_per_second": 3.981,
|
| 428 |
+
"eval_steps_per_second": 0.796,
|
| 429 |
+
"step": 4460
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 21.0,
|
| 433 |
+
"eval_loss": 4.197434425354004,
|
| 434 |
+
"eval_mean_perplexity": 552.9715087890625,
|
| 435 |
+
"eval_perplexities": [
|
| 436 |
+
496.9158630371094,
|
| 437 |
+
622.5189208984375,
|
| 438 |
+
620.3365478515625,
|
| 439 |
+
315.29119873046875,
|
| 440 |
+
508.8257751464844,
|
| 441 |
+
386.271728515625,
|
| 442 |
+
597.5241088867188,
|
| 443 |
+
540.0391845703125,
|
| 444 |
+
771.0007934570312,
|
| 445 |
+
670.990966796875
|
| 446 |
+
],
|
| 447 |
+
"eval_runtime": 2.2574,
|
| 448 |
+
"eval_samples_per_second": 4.43,
|
| 449 |
+
"eval_steps_per_second": 0.886,
|
| 450 |
+
"step": 4683
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 22.0,
|
| 454 |
+
"eval_loss": 4.2334303855896,
|
| 455 |
+
"eval_mean_perplexity": 591.004769897461,
|
| 456 |
+
"eval_perplexities": [
|
| 457 |
+
542.1947631835938,
|
| 458 |
+
699.9281616210938,
|
| 459 |
+
655.9802856445312,
|
| 460 |
+
351.9259948730469,
|
| 461 |
+
517.2227783203125,
|
| 462 |
+
364.0205078125,
|
| 463 |
+
738.9130859375,
|
| 464 |
+
537.7424926757812,
|
| 465 |
+
836.2189331054688,
|
| 466 |
+
665.9006958007812
|
| 467 |
+
],
|
| 468 |
+
"eval_runtime": 2.2478,
|
| 469 |
+
"eval_samples_per_second": 4.449,
|
| 470 |
+
"eval_steps_per_second": 0.89,
|
| 471 |
+
"step": 4906
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 22.42,
|
| 475 |
+
"grad_norm": 1.9367769956588745,
|
| 476 |
+
"learning_rate": 2.7578475336322873e-05,
|
| 477 |
+
"loss": 3.112,
|
| 478 |
+
"step": 5000
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 23.0,
|
| 482 |
+
"eval_loss": 4.230788707733154,
|
| 483 |
+
"eval_mean_perplexity": 571.9811889648438,
|
| 484 |
+
"eval_perplexities": [
|
| 485 |
+
529.1312255859375,
|
| 486 |
+
677.629638671875,
|
| 487 |
+
639.4375,
|
| 488 |
+
351.9476623535156,
|
| 489 |
+
471.0734558105469,
|
| 490 |
+
388.02935791015625,
|
| 491 |
+
630.8019409179688,
|
| 492 |
+
550.748046875,
|
| 493 |
+
854.4290771484375,
|
| 494 |
+
626.583984375
|
| 495 |
+
],
|
| 496 |
+
"eval_runtime": 2.3322,
|
| 497 |
+
"eval_samples_per_second": 4.288,
|
| 498 |
+
"eval_steps_per_second": 0.858,
|
| 499 |
+
"step": 5129
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 24.0,
|
| 503 |
+
"eval_loss": 4.264792442321777,
|
| 504 |
+
"eval_mean_perplexity": 588.0643798828125,
|
| 505 |
+
"eval_perplexities": [
|
| 506 |
+
532.3706665039062,
|
| 507 |
+
650.5995483398438,
|
| 508 |
+
640.096435546875,
|
| 509 |
+
367.4383850097656,
|
| 510 |
+
510.7347106933594,
|
| 511 |
+
402.79925537109375,
|
| 512 |
+
740.1875610351562,
|
| 513 |
+
580.5010375976562,
|
| 514 |
+
837.0123901367188,
|
| 515 |
+
618.90380859375
|
| 516 |
+
],
|
| 517 |
+
"eval_runtime": 2.414,
|
| 518 |
+
"eval_samples_per_second": 4.142,
|
| 519 |
+
"eval_steps_per_second": 0.828,
|
| 520 |
+
"step": 5352
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 25.0,
|
| 524 |
+
"eval_loss": 4.280625820159912,
|
| 525 |
+
"eval_mean_perplexity": 568.3704132080078,
|
| 526 |
+
"eval_perplexities": [
|
| 527 |
+
493.6273193359375,
|
| 528 |
+
687.0193481445312,
|
| 529 |
+
617.8013305664062,
|
| 530 |
+
356.9241943359375,
|
| 531 |
+
462.1817321777344,
|
| 532 |
+
377.38800048828125,
|
| 533 |
+
662.7510375976562,
|
| 534 |
+
553.1047973632812,
|
| 535 |
+
769.9375,
|
| 536 |
+
702.9688720703125
|
| 537 |
+
],
|
| 538 |
+
"eval_runtime": 2.2617,
|
| 539 |
+
"eval_samples_per_second": 4.421,
|
| 540 |
+
"eval_steps_per_second": 0.884,
|
| 541 |
+
"step": 5575
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 26.0,
|
| 545 |
+
"eval_loss": 4.299654006958008,
|
| 546 |
+
"eval_mean_perplexity": 585.3176055908203,
|
| 547 |
+
"eval_perplexities": [
|
| 548 |
+
507.7140808105469,
|
| 549 |
+
719.7584228515625,
|
| 550 |
+
610.5079956054688,
|
| 551 |
+
355.23577880859375,
|
| 552 |
+
474.7140197753906,
|
| 553 |
+
343.5024719238281,
|
| 554 |
+
644.6328125,
|
| 555 |
+
599.3197631835938,
|
| 556 |
+
868.6974487304688,
|
| 557 |
+
729.09326171875
|
| 558 |
+
],
|
| 559 |
+
"eval_runtime": 2.2752,
|
| 560 |
+
"eval_samples_per_second": 4.395,
|
| 561 |
+
"eval_steps_per_second": 0.879,
|
| 562 |
+
"step": 5798
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 27.0,
|
| 566 |
+
"eval_loss": 4.309383869171143,
|
| 567 |
+
"eval_mean_perplexity": 572.4837829589844,
|
| 568 |
+
"eval_perplexities": [
|
| 569 |
+
532.3777465820312,
|
| 570 |
+
701.5357055664062,
|
| 571 |
+
574.2848510742188,
|
| 572 |
+
366.85614013671875,
|
| 573 |
+
481.5206298828125,
|
| 574 |
+
389.685546875,
|
| 575 |
+
579.2503662109375,
|
| 576 |
+
525.9729614257812,
|
| 577 |
+
906.0205078125,
|
| 578 |
+
667.3333740234375
|
| 579 |
+
],
|
| 580 |
+
"eval_runtime": 2.4436,
|
| 581 |
+
"eval_samples_per_second": 4.092,
|
| 582 |
+
"eval_steps_per_second": 0.818,
|
| 583 |
+
"step": 6021
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 28.0,
|
| 587 |
+
"eval_loss": 4.338665962219238,
|
| 588 |
+
"eval_mean_perplexity": 617.5483917236328,
|
| 589 |
+
"eval_perplexities": [
|
| 590 |
+
530.51416015625,
|
| 591 |
+
770.9551391601562,
|
| 592 |
+
616.0216674804688,
|
| 593 |
+
386.1516418457031,
|
| 594 |
+
501.05426025390625,
|
| 595 |
+
419.87841796875,
|
| 596 |
+
735.8825073242188,
|
| 597 |
+
594.3335571289062,
|
| 598 |
+
911.9882202148438,
|
| 599 |
+
708.704345703125
|
| 600 |
+
],
|
| 601 |
+
"eval_runtime": 2.2597,
|
| 602 |
+
"eval_samples_per_second": 4.425,
|
| 603 |
+
"eval_steps_per_second": 0.885,
|
| 604 |
+
"step": 6244
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 29.0,
|
| 608 |
+
"eval_loss": 4.354172229766846,
|
| 609 |
+
"eval_mean_perplexity": 609.77958984375,
|
| 610 |
+
"eval_perplexities": [
|
| 611 |
+
571.1878662109375,
|
| 612 |
+
744.3867797851562,
|
| 613 |
+
607.4262084960938,
|
| 614 |
+
374.7521057128906,
|
| 615 |
+
501.3298034667969,
|
| 616 |
+
401.453369140625,
|
| 617 |
+
706.0652465820312,
|
| 618 |
+
600.0166625976562,
|
| 619 |
+
898.824951171875,
|
| 620 |
+
692.3529052734375
|
| 621 |
+
],
|
| 622 |
+
"eval_runtime": 2.4521,
|
| 623 |
+
"eval_samples_per_second": 4.078,
|
| 624 |
+
"eval_steps_per_second": 0.816,
|
| 625 |
+
"step": 6467
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 30.0,
|
| 629 |
+
"eval_loss": 4.371840000152588,
|
| 630 |
+
"eval_mean_perplexity": 639.5644592285156,
|
| 631 |
+
"eval_perplexities": [
|
| 632 |
+
518.4427490234375,
|
| 633 |
+
768.5755615234375,
|
| 634 |
+
643.536376953125,
|
| 635 |
+
434.2029724121094,
|
| 636 |
+
553.5997924804688,
|
| 637 |
+
401.9341735839844,
|
| 638 |
+
738.0169677734375,
|
| 639 |
+
627.0206909179688,
|
| 640 |
+
923.0927734375,
|
| 641 |
+
787.2225341796875
|
| 642 |
+
],
|
| 643 |
+
"eval_runtime": 2.2759,
|
| 644 |
+
"eval_samples_per_second": 4.394,
|
| 645 |
+
"eval_steps_per_second": 0.879,
|
| 646 |
+
"step": 6690
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 31.0,
|
| 650 |
+
"eval_loss": 4.397173881530762,
|
| 651 |
+
"eval_mean_perplexity": 666.7775909423829,
|
| 652 |
+
"eval_perplexities": [
|
| 653 |
+
549.1520385742188,
|
| 654 |
+
851.6005249023438,
|
| 655 |
+
712.2999267578125,
|
| 656 |
+
395.8175048828125,
|
| 657 |
+
540.697021484375,
|
| 658 |
+
451.8854064941406,
|
| 659 |
+
723.3524780273438,
|
| 660 |
+
630.9887084960938,
|
| 661 |
+
1021.4713745117188,
|
| 662 |
+
790.5109252929688
|
| 663 |
+
],
|
| 664 |
+
"eval_runtime": 2.2555,
|
| 665 |
+
"eval_samples_per_second": 4.434,
|
| 666 |
+
"eval_steps_per_second": 0.887,
|
| 667 |
+
"step": 6913
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 32.0,
|
| 671 |
+
"eval_loss": 4.415882110595703,
|
| 672 |
+
"eval_mean_perplexity": 660.8353820800781,
|
| 673 |
+
"eval_perplexities": [
|
| 674 |
+
512.7852783203125,
|
| 675 |
+
785.50439453125,
|
| 676 |
+
668.9815063476562,
|
| 677 |
+
457.69830322265625,
|
| 678 |
+
539.0671997070312,
|
| 679 |
+
459.60736083984375,
|
| 680 |
+
751.9481201171875,
|
| 681 |
+
625.0054321289062,
|
| 682 |
+
999.4456787109375,
|
| 683 |
+
808.310546875
|
| 684 |
+
],
|
| 685 |
+
"eval_runtime": 2.2544,
|
| 686 |
+
"eval_samples_per_second": 4.436,
|
| 687 |
+
"eval_steps_per_second": 0.887,
|
| 688 |
+
"step": 7136
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 33.0,
|
| 692 |
+
"eval_loss": 4.42371940612793,
|
| 693 |
+
"eval_mean_perplexity": 662.7573822021484,
|
| 694 |
+
"eval_perplexities": [
|
| 695 |
+
559.4247436523438,
|
| 696 |
+
747.0921020507812,
|
| 697 |
+
678.9531860351562,
|
| 698 |
+
437.0193786621094,
|
| 699 |
+
544.0770874023438,
|
| 700 |
+
431.26568603515625,
|
| 701 |
+
756.7093505859375,
|
| 702 |
+
626.279052734375,
|
| 703 |
+
1022.3748168945312,
|
| 704 |
+
824.37841796875
|
| 705 |
+
],
|
| 706 |
+
"eval_runtime": 2.2901,
|
| 707 |
+
"eval_samples_per_second": 4.367,
|
| 708 |
+
"eval_steps_per_second": 0.873,
|
| 709 |
+
"step": 7359
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 34.0,
|
| 713 |
+
"eval_loss": 4.434357166290283,
|
| 714 |
+
"eval_mean_perplexity": 650.7052612304688,
|
| 715 |
+
"eval_perplexities": [
|
| 716 |
+
549.6817626953125,
|
| 717 |
+
811.8685302734375,
|
| 718 |
+
639.683349609375,
|
| 719 |
+
410.87249755859375,
|
| 720 |
+
510.8334045410156,
|
| 721 |
+
438.6333923339844,
|
| 722 |
+
779.5153198242188,
|
| 723 |
+
584.6727294921875,
|
| 724 |
+
1027.2081298828125,
|
| 725 |
+
754.08349609375
|
| 726 |
+
],
|
| 727 |
+
"eval_runtime": 2.3566,
|
| 728 |
+
"eval_samples_per_second": 4.243,
|
| 729 |
+
"eval_steps_per_second": 0.849,
|
| 730 |
+
"step": 7582
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 35.0,
|
| 734 |
+
"eval_loss": 4.456236839294434,
|
| 735 |
+
"eval_mean_perplexity": 683.213168334961,
|
| 736 |
+
"eval_perplexities": [
|
| 737 |
+
584.2586059570312,
|
| 738 |
+
837.6990966796875,
|
| 739 |
+
639.265869140625,
|
| 740 |
+
431.7611389160156,
|
| 741 |
+
558.5822143554688,
|
| 742 |
+
446.50067138671875,
|
| 743 |
+
781.8605346679688,
|
| 744 |
+
649.52392578125,
|
| 745 |
+
1031.22900390625,
|
| 746 |
+
871.4506225585938
|
| 747 |
+
],
|
| 748 |
+
"eval_runtime": 2.2758,
|
| 749 |
+
"eval_samples_per_second": 4.394,
|
| 750 |
+
"eval_steps_per_second": 0.879,
|
| 751 |
+
"step": 7805
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 36.0,
|
| 755 |
+
"eval_loss": 4.476213455200195,
|
| 756 |
+
"eval_mean_perplexity": 673.9735595703125,
|
| 757 |
+
"eval_perplexities": [
|
| 758 |
+
570.093505859375,
|
| 759 |
+
876.3085327148438,
|
| 760 |
+
637.6167602539062,
|
| 761 |
+
410.8495788574219,
|
| 762 |
+
565.6197509765625,
|
| 763 |
+
431.1340637207031,
|
| 764 |
+
784.1497192382812,
|
| 765 |
+
635.3842163085938,
|
| 766 |
+
1020.8198852539062,
|
| 767 |
+
807.7595825195312
|
| 768 |
+
],
|
| 769 |
+
"eval_runtime": 2.2666,
|
| 770 |
+
"eval_samples_per_second": 4.412,
|
| 771 |
+
"eval_steps_per_second": 0.882,
|
| 772 |
+
"step": 8028
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 37.0,
|
| 776 |
+
"eval_loss": 4.485024452209473,
|
| 777 |
+
"eval_mean_perplexity": 678.1103912353516,
|
| 778 |
+
"eval_perplexities": [
|
| 779 |
+
524.7376708984375,
|
| 780 |
+
897.5795288085938,
|
| 781 |
+
604.8717651367188,
|
| 782 |
+
421.68621826171875,
|
| 783 |
+
562.4842529296875,
|
| 784 |
+
447.8833312988281,
|
| 785 |
+
743.5353393554688,
|
| 786 |
+
641.0125122070312,
|
| 787 |
+
1038.24658203125,
|
| 788 |
+
899.0667114257812
|
| 789 |
+
],
|
| 790 |
+
"eval_runtime": 2.2714,
|
| 791 |
+
"eval_samples_per_second": 4.403,
|
| 792 |
+
"eval_steps_per_second": 0.881,
|
| 793 |
+
"step": 8251
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 38.0,
|
| 797 |
+
"eval_loss": 4.500662326812744,
|
| 798 |
+
"eval_mean_perplexity": 667.2204986572266,
|
| 799 |
+
"eval_perplexities": [
|
| 800 |
+
517.3828125,
|
| 801 |
+
849.9619140625,
|
| 802 |
+
632.2133178710938,
|
| 803 |
+
421.2346496582031,
|
| 804 |
+
562.2227783203125,
|
| 805 |
+
412.367919921875,
|
| 806 |
+
801.9597778320312,
|
| 807 |
+
634.914794921875,
|
| 808 |
+
955.3936767578125,
|
| 809 |
+
884.5533447265625
|
| 810 |
+
],
|
| 811 |
+
"eval_runtime": 2.2586,
|
| 812 |
+
"eval_samples_per_second": 4.428,
|
| 813 |
+
"eval_steps_per_second": 0.886,
|
| 814 |
+
"step": 8474
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 39.0,
|
| 818 |
+
"eval_loss": 4.507510662078857,
|
| 819 |
+
"eval_mean_perplexity": 674.6702362060547,
|
| 820 |
+
"eval_perplexities": [
|
| 821 |
+
523.6810913085938,
|
| 822 |
+
871.532958984375,
|
| 823 |
+
702.7888793945312,
|
| 824 |
+
412.113525390625,
|
| 825 |
+
559.32421875,
|
| 826 |
+
436.3467712402344,
|
| 827 |
+
741.5978393554688,
|
| 828 |
+
612.4104614257812,
|
| 829 |
+
969.3262329101562,
|
| 830 |
+
917.5803833007812
|
| 831 |
+
],
|
| 832 |
+
"eval_runtime": 2.2926,
|
| 833 |
+
"eval_samples_per_second": 4.362,
|
| 834 |
+
"eval_steps_per_second": 0.872,
|
| 835 |
+
"step": 8697
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 40.0,
|
| 839 |
+
"eval_loss": 4.5184006690979,
|
| 840 |
+
"eval_mean_perplexity": 690.1411804199219,
|
| 841 |
+
"eval_perplexities": [
|
| 842 |
+
533.463623046875,
|
| 843 |
+
880.5369262695312,
|
| 844 |
+
677.1242065429688,
|
| 845 |
+
468.8255920410156,
|
| 846 |
+
524.9594116210938,
|
| 847 |
+
481.3450012207031,
|
| 848 |
+
724.6446533203125,
|
| 849 |
+
644.5559692382812,
|
| 850 |
+
1083.451416015625,
|
| 851 |
+
882.5050048828125
|
| 852 |
+
],
|
| 853 |
+
"eval_runtime": 2.2507,
|
| 854 |
+
"eval_samples_per_second": 4.443,
|
| 855 |
+
"eval_steps_per_second": 0.889,
|
| 856 |
+
"step": 8920
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 41.0,
|
| 860 |
+
"eval_loss": 4.528339385986328,
|
| 861 |
+
"eval_mean_perplexity": 686.3480163574219,
|
| 862 |
+
"eval_perplexities": [
|
| 863 |
+
540.8236083984375,
|
| 864 |
+
941.376708984375,
|
| 865 |
+
641.9276733398438,
|
| 866 |
+
446.772216796875,
|
| 867 |
+
562.5877685546875,
|
| 868 |
+
455.73175048828125,
|
| 869 |
+
764.4960327148438,
|
| 870 |
+
647.960205078125,
|
| 871 |
+
983.826416015625,
|
| 872 |
+
877.977783203125
|
| 873 |
+
],
|
| 874 |
+
"eval_runtime": 2.2901,
|
| 875 |
+
"eval_samples_per_second": 4.367,
|
| 876 |
+
"eval_steps_per_second": 0.873,
|
| 877 |
+
"step": 9143
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 42.0,
|
| 881 |
+
"eval_loss": 4.525776386260986,
|
| 882 |
+
"eval_mean_perplexity": 692.94345703125,
|
| 883 |
+
"eval_perplexities": [
|
| 884 |
+
512.2537841796875,
|
| 885 |
+
996.9277954101562,
|
| 886 |
+
626.796142578125,
|
| 887 |
+
407.5407409667969,
|
| 888 |
+
572.6943969726562,
|
| 889 |
+
439.5016174316406,
|
| 890 |
+
776.8684692382812,
|
| 891 |
+
660.894775390625,
|
| 892 |
+
1015.8809204101562,
|
| 893 |
+
920.075927734375
|
| 894 |
+
],
|
| 895 |
+
"eval_runtime": 2.2958,
|
| 896 |
+
"eval_samples_per_second": 4.356,
|
| 897 |
+
"eval_steps_per_second": 0.871,
|
| 898 |
+
"step": 9366
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 43.0,
|
| 902 |
+
"eval_loss": 4.5432844161987305,
|
| 903 |
+
"eval_mean_perplexity": 698.305697631836,
|
| 904 |
+
"eval_perplexities": [
|
| 905 |
+
520.623779296875,
|
| 906 |
+
898.4200439453125,
|
| 907 |
+
635.4287719726562,
|
| 908 |
+
451.1359558105469,
|
| 909 |
+
590.0744018554688,
|
| 910 |
+
448.103759765625,
|
| 911 |
+
752.5787353515625,
|
| 912 |
+
682.3783569335938,
|
| 913 |
+
1070.27294921875,
|
| 914 |
+
934.0402221679688
|
| 915 |
+
],
|
| 916 |
+
"eval_runtime": 2.8416,
|
| 917 |
+
"eval_samples_per_second": 3.519,
|
| 918 |
+
"eval_steps_per_second": 0.704,
|
| 919 |
+
"step": 9589
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 44.0,
|
| 923 |
+
"eval_loss": 4.546627998352051,
|
| 924 |
+
"eval_mean_perplexity": 698.2709533691407,
|
| 925 |
+
"eval_perplexities": [
|
| 926 |
+
496.5508728027344,
|
| 927 |
+
900.9185180664062,
|
| 928 |
+
651.6212768554688,
|
| 929 |
+
429.4538879394531,
|
| 930 |
+
572.0066528320312,
|
| 931 |
+
429.689208984375,
|
| 932 |
+
786.3671264648438,
|
| 933 |
+
686.788818359375,
|
| 934 |
+
1078.3935546875,
|
| 935 |
+
950.9196166992188
|
| 936 |
+
],
|
| 937 |
+
"eval_runtime": 2.2997,
|
| 938 |
+
"eval_samples_per_second": 4.348,
|
| 939 |
+
"eval_steps_per_second": 0.87,
|
| 940 |
+
"step": 9812
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 44.84,
|
| 944 |
+
"grad_norm": 2.2081024646759033,
|
| 945 |
+
"learning_rate": 5.15695067264574e-06,
|
| 946 |
+
"loss": 2.4732,
|
| 947 |
+
"step": 10000
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 45.0,
|
| 951 |
+
"eval_loss": 4.5482940673828125,
|
| 952 |
+
"eval_mean_perplexity": 683.0953948974609,
|
| 953 |
+
"eval_perplexities": [
|
| 954 |
+
513.7635498046875,
|
| 955 |
+
881.8916625976562,
|
| 956 |
+
650.1325073242188,
|
| 957 |
+
426.90740966796875,
|
| 958 |
+
583.7874145507812,
|
| 959 |
+
441.8877258300781,
|
| 960 |
+
793.0048828125,
|
| 961 |
+
660.0620727539062,
|
| 962 |
+
992.5838012695312,
|
| 963 |
+
886.9329223632812
|
| 964 |
+
],
|
| 965 |
+
"eval_runtime": 2.6736,
|
| 966 |
+
"eval_samples_per_second": 3.74,
|
| 967 |
+
"eval_steps_per_second": 0.748,
|
| 968 |
+
"step": 10035
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 46.0,
|
| 972 |
+
"eval_loss": 4.556254863739014,
|
| 973 |
+
"eval_mean_perplexity": 694.8246520996094,
|
| 974 |
+
"eval_perplexities": [
|
| 975 |
+
484.3752136230469,
|
| 976 |
+
864.9951171875,
|
| 977 |
+
721.4644775390625,
|
| 978 |
+
440.7280578613281,
|
| 979 |
+
583.1522216796875,
|
| 980 |
+
458.59246826171875,
|
| 981 |
+
795.0460815429688,
|
| 982 |
+
658.3065795898438,
|
| 983 |
+
1043.74267578125,
|
| 984 |
+
897.8436279296875
|
| 985 |
+
],
|
| 986 |
+
"eval_runtime": 2.6592,
|
| 987 |
+
"eval_samples_per_second": 3.761,
|
| 988 |
+
"eval_steps_per_second": 0.752,
|
| 989 |
+
"step": 10258
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 47.0,
|
| 993 |
+
"eval_loss": 4.565934181213379,
|
| 994 |
+
"eval_mean_perplexity": 697.1294586181641,
|
| 995 |
+
"eval_perplexities": [
|
| 996 |
+
521.9780883789062,
|
| 997 |
+
891.4306030273438,
|
| 998 |
+
694.7803344726562,
|
| 999 |
+
465.2270812988281,
|
| 1000 |
+
585.0350341796875,
|
| 1001 |
+
454.22308349609375,
|
| 1002 |
+
784.2135620117188,
|
| 1003 |
+
672.1557006835938,
|
| 1004 |
+
1016.0956420898438,
|
| 1005 |
+
886.1554565429688
|
| 1006 |
+
],
|
| 1007 |
+
"eval_runtime": 2.3829,
|
| 1008 |
+
"eval_samples_per_second": 4.197,
|
| 1009 |
+
"eval_steps_per_second": 0.839,
|
| 1010 |
+
"step": 10481
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 48.0,
|
| 1014 |
+
"eval_loss": 4.569299221038818,
|
| 1015 |
+
"eval_mean_perplexity": 698.8598907470703,
|
| 1016 |
+
"eval_perplexities": [
|
| 1017 |
+
523.372802734375,
|
| 1018 |
+
930.4688720703125,
|
| 1019 |
+
683.875732421875,
|
| 1020 |
+
451.5999755859375,
|
| 1021 |
+
597.7974243164062,
|
| 1022 |
+
447.0304870605469,
|
| 1023 |
+
766.9913940429688,
|
| 1024 |
+
671.3782348632812,
|
| 1025 |
+
1039.4984130859375,
|
| 1026 |
+
876.5855712890625
|
| 1027 |
+
],
|
| 1028 |
+
"eval_runtime": 2.2797,
|
| 1029 |
+
"eval_samples_per_second": 4.387,
|
| 1030 |
+
"eval_steps_per_second": 0.877,
|
| 1031 |
+
"step": 10704
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 49.0,
|
| 1035 |
+
"eval_loss": 4.569689750671387,
|
| 1036 |
+
"eval_mean_perplexity": 694.5519165039062,
|
| 1037 |
+
"eval_perplexities": [
|
| 1038 |
+
508.9573059082031,
|
| 1039 |
+
927.446533203125,
|
| 1040 |
+
672.3489379882812,
|
| 1041 |
+
473.5368347167969,
|
| 1042 |
+
568.0377807617188,
|
| 1043 |
+
454.00933837890625,
|
| 1044 |
+
733.6597900390625,
|
| 1045 |
+
695.4989624023438,
|
| 1046 |
+
1047.673828125,
|
| 1047 |
+
864.349853515625
|
| 1048 |
+
],
|
| 1049 |
+
"eval_runtime": 2.3973,
|
| 1050 |
+
"eval_samples_per_second": 4.171,
|
| 1051 |
+
"eval_steps_per_second": 0.834,
|
| 1052 |
+
"step": 10927
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 50.0,
|
| 1056 |
+
"eval_loss": 4.569555759429932,
|
| 1057 |
+
"eval_mean_perplexity": 698.6035186767579,
|
| 1058 |
+
"eval_perplexities": [
|
| 1059 |
+
517.4149169921875,
|
| 1060 |
+
924.535888671875,
|
| 1061 |
+
704.73291015625,
|
| 1062 |
+
465.9677429199219,
|
| 1063 |
+
577.629150390625,
|
| 1064 |
+
443.994140625,
|
| 1065 |
+
770.1861572265625,
|
| 1066 |
+
683.028076171875,
|
| 1067 |
+
1017.7510375976562,
|
| 1068 |
+
880.795166015625
|
| 1069 |
+
],
|
| 1070 |
+
"eval_runtime": 2.2771,
|
| 1071 |
+
"eval_samples_per_second": 4.391,
|
| 1072 |
+
"eval_steps_per_second": 0.878,
|
| 1073 |
+
"step": 11150
|
| 1074 |
+
}
|
| 1075 |
+
],
|
| 1076 |
+
"logging_steps": 5000,
|
| 1077 |
+
"max_steps": 11150,
|
| 1078 |
+
"num_input_tokens_seen": 0,
|
| 1079 |
+
"num_train_epochs": 50,
|
| 1080 |
+
"save_steps": 500,
|
| 1081 |
+
"total_flos": 1.1355181056e+16,
|
| 1082 |
+
"train_batch_size": 10,
|
| 1083 |
+
"trial_name": null,
|
| 1084 |
+
"trial_params": null
|
| 1085 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6ebdc821e3d628e3bfbc7af5db7645e4fcff47fe3fbf10447699dcc928e2825d
|
| 3 |
+
size 4920
|