Upload 8 files
Browse filesadding the whole checkpoint folder (with optimizer, model, config etc) for the gpt2 small-like 12 layer model
- config.json +30 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- trainer_state.json +822 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_function": "gelu_new",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"GPT2LMHeadModel"
|
| 5 |
+
],
|
| 6 |
+
"attn_pdrop": 0.1,
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"embd_pdrop": 0.1,
|
| 9 |
+
"eos_token_id": 0,
|
| 10 |
+
"gradient_checkpointing": false,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"layer_norm_epsilon": 1e-05,
|
| 13 |
+
"model_type": "gpt2",
|
| 14 |
+
"n_ctx": 1024,
|
| 15 |
+
"n_embd": 768,
|
| 16 |
+
"n_head": 12,
|
| 17 |
+
"n_inner": null,
|
| 18 |
+
"n_layer": 12,
|
| 19 |
+
"n_positions": 1024,
|
| 20 |
+
"resid_pdrop": 0.1,
|
| 21 |
+
"scale_attn_weights": true,
|
| 22 |
+
"summary_activation": null,
|
| 23 |
+
"summary_first_dropout": 0.1,
|
| 24 |
+
"summary_proj_to_labels": true,
|
| 25 |
+
"summary_type": "cls_index",
|
| 26 |
+
"summary_use_proj": true,
|
| 27 |
+
"transformers_version": "4.6.1",
|
| 28 |
+
"use_cache": true,
|
| 29 |
+
"vocab_size": 28439
|
| 30 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d5fc39bca73ec237baf3ac9a798f6440424e2563b4119332e70464a05e0b04fe
|
| 3 |
+
size 861561495
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:275fb18cbd8fe5dbb8408286abf41616807dc35d954fda3770642866f22d50cb
|
| 3 |
+
size 443383419
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb6ce2e55d10782731f1ba5fa76fd3a0aaad0a59a0fa74b67e588a896b35eac1
|
| 3 |
+
size 14657
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:923ead5faf30bd8c7efdc44ed28f0b833c8fceb006bd42e6c034f8faf6fd4ec1
|
| 3 |
+
size 559
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b9dc201b0daadb61ba308d04c165a144a35adf1b3f4de97f9d3ad94c0e36d533
|
| 3 |
+
size 559
|
trainer_state.json
ADDED
|
@@ -0,0 +1,822 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 3.699657678604126,
|
| 3 |
+
"best_model_checkpoint": "/scratch/ka2773/project/lm-mem/checkpoints/gpt2_40m_12-768-1024_a_02/checkpoint-31000",
|
| 4 |
+
"epoch": 8.481532147742818,
|
| 5 |
+
"global_step": 31000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.14,
|
| 12 |
+
"learning_rate": 6e-05,
|
| 13 |
+
"loss": 6.6803,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.14,
|
| 18 |
+
"eval_loss": 6.017919540405273,
|
| 19 |
+
"eval_runtime": 9.4636,
|
| 20 |
+
"eval_samples_per_second": 24.938,
|
| 21 |
+
"step": 500
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.27,
|
| 25 |
+
"learning_rate": 6e-05,
|
| 26 |
+
"loss": 6.0117,
|
| 27 |
+
"step": 1000
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.27,
|
| 31 |
+
"eval_loss": 5.752170085906982,
|
| 32 |
+
"eval_runtime": 9.4619,
|
| 33 |
+
"eval_samples_per_second": 24.942,
|
| 34 |
+
"step": 1000
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 0.41,
|
| 38 |
+
"learning_rate": 6e-05,
|
| 39 |
+
"loss": 5.8304,
|
| 40 |
+
"step": 1500
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.41,
|
| 44 |
+
"eval_loss": 5.598978519439697,
|
| 45 |
+
"eval_runtime": 9.4601,
|
| 46 |
+
"eval_samples_per_second": 24.947,
|
| 47 |
+
"step": 1500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.55,
|
| 51 |
+
"learning_rate": 6e-05,
|
| 52 |
+
"loss": 5.6932,
|
| 53 |
+
"step": 2000
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.55,
|
| 57 |
+
"eval_loss": 5.4606218338012695,
|
| 58 |
+
"eval_runtime": 9.4514,
|
| 59 |
+
"eval_samples_per_second": 24.97,
|
| 60 |
+
"step": 2000
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.68,
|
| 64 |
+
"learning_rate": 6e-05,
|
| 65 |
+
"loss": 5.5582,
|
| 66 |
+
"step": 2500
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.68,
|
| 70 |
+
"eval_loss": 5.324854373931885,
|
| 71 |
+
"eval_runtime": 9.4555,
|
| 72 |
+
"eval_samples_per_second": 24.959,
|
| 73 |
+
"step": 2500
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.82,
|
| 77 |
+
"learning_rate": 6e-05,
|
| 78 |
+
"loss": 5.4194,
|
| 79 |
+
"step": 3000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.82,
|
| 83 |
+
"eval_loss": 5.187784194946289,
|
| 84 |
+
"eval_runtime": 9.4406,
|
| 85 |
+
"eval_samples_per_second": 24.998,
|
| 86 |
+
"step": 3000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.96,
|
| 90 |
+
"learning_rate": 6e-05,
|
| 91 |
+
"loss": 5.2874,
|
| 92 |
+
"step": 3500
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.96,
|
| 96 |
+
"eval_loss": 5.043916702270508,
|
| 97 |
+
"eval_runtime": 9.4594,
|
| 98 |
+
"eval_samples_per_second": 24.949,
|
| 99 |
+
"step": 3500
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 1.09,
|
| 103 |
+
"learning_rate": 6e-05,
|
| 104 |
+
"loss": 5.1502,
|
| 105 |
+
"step": 4000
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.09,
|
| 109 |
+
"eval_loss": 4.9315690994262695,
|
| 110 |
+
"eval_runtime": 9.458,
|
| 111 |
+
"eval_samples_per_second": 24.952,
|
| 112 |
+
"step": 4000
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.23,
|
| 116 |
+
"learning_rate": 6e-05,
|
| 117 |
+
"loss": 5.0351,
|
| 118 |
+
"step": 4500
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 1.23,
|
| 122 |
+
"eval_loss": 4.828794479370117,
|
| 123 |
+
"eval_runtime": 9.4563,
|
| 124 |
+
"eval_samples_per_second": 24.957,
|
| 125 |
+
"step": 4500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.37,
|
| 129 |
+
"learning_rate": 6e-05,
|
| 130 |
+
"loss": 4.9495,
|
| 131 |
+
"step": 5000
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.37,
|
| 135 |
+
"eval_loss": 4.735019683837891,
|
| 136 |
+
"eval_runtime": 9.4648,
|
| 137 |
+
"eval_samples_per_second": 24.934,
|
| 138 |
+
"step": 5000
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 1.5,
|
| 142 |
+
"learning_rate": 6e-05,
|
| 143 |
+
"loss": 4.8715,
|
| 144 |
+
"step": 5500
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.5,
|
| 148 |
+
"eval_loss": 4.6624579429626465,
|
| 149 |
+
"eval_runtime": 9.4463,
|
| 150 |
+
"eval_samples_per_second": 24.983,
|
| 151 |
+
"step": 5500
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.64,
|
| 155 |
+
"learning_rate": 6e-05,
|
| 156 |
+
"loss": 4.7962,
|
| 157 |
+
"step": 6000
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 1.64,
|
| 161 |
+
"eval_loss": 4.586391925811768,
|
| 162 |
+
"eval_runtime": 9.4563,
|
| 163 |
+
"eval_samples_per_second": 24.957,
|
| 164 |
+
"step": 6000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.78,
|
| 168 |
+
"learning_rate": 6e-05,
|
| 169 |
+
"loss": 4.7329,
|
| 170 |
+
"step": 6500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.78,
|
| 174 |
+
"eval_loss": 4.520113468170166,
|
| 175 |
+
"eval_runtime": 9.4613,
|
| 176 |
+
"eval_samples_per_second": 24.944,
|
| 177 |
+
"step": 6500
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 1.92,
|
| 181 |
+
"learning_rate": 6e-05,
|
| 182 |
+
"loss": 4.6692,
|
| 183 |
+
"step": 7000
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 1.92,
|
| 187 |
+
"eval_loss": 4.469329357147217,
|
| 188 |
+
"eval_runtime": 9.4531,
|
| 189 |
+
"eval_samples_per_second": 24.965,
|
| 190 |
+
"step": 7000
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 2.05,
|
| 194 |
+
"learning_rate": 6e-05,
|
| 195 |
+
"loss": 4.603,
|
| 196 |
+
"step": 7500
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 2.05,
|
| 200 |
+
"eval_loss": 4.408419132232666,
|
| 201 |
+
"eval_runtime": 9.4573,
|
| 202 |
+
"eval_samples_per_second": 24.954,
|
| 203 |
+
"step": 7500
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 2.19,
|
| 207 |
+
"learning_rate": 6e-05,
|
| 208 |
+
"loss": 4.5274,
|
| 209 |
+
"step": 8000
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.19,
|
| 213 |
+
"eval_loss": 4.358486175537109,
|
| 214 |
+
"eval_runtime": 9.4556,
|
| 215 |
+
"eval_samples_per_second": 24.959,
|
| 216 |
+
"step": 8000
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.33,
|
| 220 |
+
"learning_rate": 6e-05,
|
| 221 |
+
"loss": 4.482,
|
| 222 |
+
"step": 8500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 2.33,
|
| 226 |
+
"eval_loss": 4.316709041595459,
|
| 227 |
+
"eval_runtime": 9.4572,
|
| 228 |
+
"eval_samples_per_second": 24.954,
|
| 229 |
+
"step": 8500
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 2.46,
|
| 233 |
+
"learning_rate": 6e-05,
|
| 234 |
+
"loss": 4.4431,
|
| 235 |
+
"step": 9000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 2.46,
|
| 239 |
+
"eval_loss": 4.261863708496094,
|
| 240 |
+
"eval_runtime": 9.4644,
|
| 241 |
+
"eval_samples_per_second": 24.935,
|
| 242 |
+
"step": 9000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.6,
|
| 246 |
+
"learning_rate": 6e-05,
|
| 247 |
+
"loss": 4.4003,
|
| 248 |
+
"step": 9500
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.6,
|
| 252 |
+
"eval_loss": 4.2196125984191895,
|
| 253 |
+
"eval_runtime": 9.4647,
|
| 254 |
+
"eval_samples_per_second": 24.935,
|
| 255 |
+
"step": 9500
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 2.74,
|
| 259 |
+
"learning_rate": 6e-05,
|
| 260 |
+
"loss": 4.3602,
|
| 261 |
+
"step": 10000
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 2.74,
|
| 265 |
+
"eval_loss": 4.1680402755737305,
|
| 266 |
+
"eval_runtime": 9.4399,
|
| 267 |
+
"eval_samples_per_second": 25.0,
|
| 268 |
+
"step": 10000
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 2.87,
|
| 272 |
+
"learning_rate": 6e-05,
|
| 273 |
+
"loss": 4.3072,
|
| 274 |
+
"step": 10500
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 2.87,
|
| 278 |
+
"eval_loss": 4.134276390075684,
|
| 279 |
+
"eval_runtime": 9.4305,
|
| 280 |
+
"eval_samples_per_second": 25.025,
|
| 281 |
+
"step": 10500
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 3.01,
|
| 285 |
+
"learning_rate": 6e-05,
|
| 286 |
+
"loss": 4.2849,
|
| 287 |
+
"step": 11000
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 3.01,
|
| 291 |
+
"eval_loss": 4.093378067016602,
|
| 292 |
+
"eval_runtime": 9.4676,
|
| 293 |
+
"eval_samples_per_second": 24.927,
|
| 294 |
+
"step": 11000
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 3.15,
|
| 298 |
+
"learning_rate": 6e-05,
|
| 299 |
+
"loss": 4.2092,
|
| 300 |
+
"step": 11500
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 3.15,
|
| 304 |
+
"eval_loss": 4.067226886749268,
|
| 305 |
+
"eval_runtime": 9.4548,
|
| 306 |
+
"eval_samples_per_second": 24.961,
|
| 307 |
+
"step": 11500
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 3.28,
|
| 311 |
+
"learning_rate": 6e-05,
|
| 312 |
+
"loss": 4.1923,
|
| 313 |
+
"step": 12000
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 3.28,
|
| 317 |
+
"eval_loss": 4.044227600097656,
|
| 318 |
+
"eval_runtime": 9.4733,
|
| 319 |
+
"eval_samples_per_second": 24.912,
|
| 320 |
+
"step": 12000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 3.42,
|
| 324 |
+
"learning_rate": 6e-05,
|
| 325 |
+
"loss": 4.1774,
|
| 326 |
+
"step": 12500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 3.42,
|
| 330 |
+
"eval_loss": 4.027223587036133,
|
| 331 |
+
"eval_runtime": 9.4672,
|
| 332 |
+
"eval_samples_per_second": 24.928,
|
| 333 |
+
"step": 12500
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 3.56,
|
| 337 |
+
"learning_rate": 6e-05,
|
| 338 |
+
"loss": 4.1547,
|
| 339 |
+
"step": 13000
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 3.56,
|
| 343 |
+
"eval_loss": 4.007537841796875,
|
| 344 |
+
"eval_runtime": 9.4577,
|
| 345 |
+
"eval_samples_per_second": 24.953,
|
| 346 |
+
"step": 13000
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 3.69,
|
| 350 |
+
"learning_rate": 6e-05,
|
| 351 |
+
"loss": 4.141,
|
| 352 |
+
"step": 13500
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 3.69,
|
| 356 |
+
"eval_loss": 3.9847309589385986,
|
| 357 |
+
"eval_runtime": 9.464,
|
| 358 |
+
"eval_samples_per_second": 24.937,
|
| 359 |
+
"step": 13500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.83,
|
| 363 |
+
"learning_rate": 6e-05,
|
| 364 |
+
"loss": 4.1375,
|
| 365 |
+
"step": 14000
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.83,
|
| 369 |
+
"eval_loss": 3.9671761989593506,
|
| 370 |
+
"eval_runtime": 9.4732,
|
| 371 |
+
"eval_samples_per_second": 24.912,
|
| 372 |
+
"step": 14000
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 3.97,
|
| 376 |
+
"learning_rate": 6e-05,
|
| 377 |
+
"loss": 4.1064,
|
| 378 |
+
"step": 14500
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 3.97,
|
| 382 |
+
"eval_loss": 3.9399986267089844,
|
| 383 |
+
"eval_runtime": 9.4647,
|
| 384 |
+
"eval_samples_per_second": 24.935,
|
| 385 |
+
"step": 14500
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 4.1,
|
| 389 |
+
"learning_rate": 6e-05,
|
| 390 |
+
"loss": 4.0485,
|
| 391 |
+
"step": 15000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 4.1,
|
| 395 |
+
"eval_loss": 3.932234525680542,
|
| 396 |
+
"eval_runtime": 9.4672,
|
| 397 |
+
"eval_samples_per_second": 24.928,
|
| 398 |
+
"step": 15000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 4.24,
|
| 402 |
+
"learning_rate": 6e-05,
|
| 403 |
+
"loss": 4.0335,
|
| 404 |
+
"step": 15500
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 4.24,
|
| 408 |
+
"eval_loss": 3.9215309619903564,
|
| 409 |
+
"eval_runtime": 9.4597,
|
| 410 |
+
"eval_samples_per_second": 24.948,
|
| 411 |
+
"step": 15500
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 4.38,
|
| 415 |
+
"learning_rate": 6e-05,
|
| 416 |
+
"loss": 4.0358,
|
| 417 |
+
"step": 16000
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 4.38,
|
| 421 |
+
"eval_loss": 3.898069381713867,
|
| 422 |
+
"eval_runtime": 9.4635,
|
| 423 |
+
"eval_samples_per_second": 24.938,
|
| 424 |
+
"step": 16000
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 4.51,
|
| 428 |
+
"learning_rate": 6e-05,
|
| 429 |
+
"loss": 4.0187,
|
| 430 |
+
"step": 16500
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 4.51,
|
| 434 |
+
"eval_loss": 3.893036127090454,
|
| 435 |
+
"eval_runtime": 9.4622,
|
| 436 |
+
"eval_samples_per_second": 24.941,
|
| 437 |
+
"step": 16500
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 4.65,
|
| 441 |
+
"learning_rate": 6e-05,
|
| 442 |
+
"loss": 4.0188,
|
| 443 |
+
"step": 17000
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 4.65,
|
| 447 |
+
"eval_loss": 3.8782527446746826,
|
| 448 |
+
"eval_runtime": 9.459,
|
| 449 |
+
"eval_samples_per_second": 24.95,
|
| 450 |
+
"step": 17000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 4.79,
|
| 454 |
+
"learning_rate": 6e-05,
|
| 455 |
+
"loss": 4.0176,
|
| 456 |
+
"step": 17500
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 4.79,
|
| 460 |
+
"eval_loss": 3.8655033111572266,
|
| 461 |
+
"eval_runtime": 9.4578,
|
| 462 |
+
"eval_samples_per_second": 24.953,
|
| 463 |
+
"step": 17500
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 4.92,
|
| 467 |
+
"learning_rate": 6e-05,
|
| 468 |
+
"loss": 4.0025,
|
| 469 |
+
"step": 18000
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 4.92,
|
| 473 |
+
"eval_loss": 3.8568382263183594,
|
| 474 |
+
"eval_runtime": 9.4581,
|
| 475 |
+
"eval_samples_per_second": 24.952,
|
| 476 |
+
"step": 18000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 5.06,
|
| 480 |
+
"learning_rate": 6e-05,
|
| 481 |
+
"loss": 3.9554,
|
| 482 |
+
"step": 18500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 5.06,
|
| 486 |
+
"eval_loss": 3.849705696105957,
|
| 487 |
+
"eval_runtime": 9.4471,
|
| 488 |
+
"eval_samples_per_second": 24.981,
|
| 489 |
+
"step": 18500
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 5.2,
|
| 493 |
+
"learning_rate": 6e-05,
|
| 494 |
+
"loss": 3.9252,
|
| 495 |
+
"step": 19000
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 5.2,
|
| 499 |
+
"eval_loss": 3.838930130004883,
|
| 500 |
+
"eval_runtime": 9.4605,
|
| 501 |
+
"eval_samples_per_second": 24.946,
|
| 502 |
+
"step": 19000
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 5.34,
|
| 506 |
+
"learning_rate": 6e-05,
|
| 507 |
+
"loss": 3.9239,
|
| 508 |
+
"step": 19500
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 5.34,
|
| 512 |
+
"eval_loss": 3.8333945274353027,
|
| 513 |
+
"eval_runtime": 9.462,
|
| 514 |
+
"eval_samples_per_second": 24.942,
|
| 515 |
+
"step": 19500
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 5.47,
|
| 519 |
+
"learning_rate": 6e-05,
|
| 520 |
+
"loss": 3.9354,
|
| 521 |
+
"step": 20000
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 5.47,
|
| 525 |
+
"eval_loss": 3.8130805492401123,
|
| 526 |
+
"eval_runtime": 9.4596,
|
| 527 |
+
"eval_samples_per_second": 24.948,
|
| 528 |
+
"step": 20000
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 5.61,
|
| 532 |
+
"learning_rate": 6e-05,
|
| 533 |
+
"loss": 3.9418,
|
| 534 |
+
"step": 20500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 5.61,
|
| 538 |
+
"eval_loss": 3.8175787925720215,
|
| 539 |
+
"eval_runtime": 9.4602,
|
| 540 |
+
"eval_samples_per_second": 24.947,
|
| 541 |
+
"step": 20500
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 5.75,
|
| 545 |
+
"learning_rate": 6e-05,
|
| 546 |
+
"loss": 3.9291,
|
| 547 |
+
"step": 21000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 5.75,
|
| 551 |
+
"eval_loss": 3.807772636413574,
|
| 552 |
+
"eval_runtime": 9.459,
|
| 553 |
+
"eval_samples_per_second": 24.95,
|
| 554 |
+
"step": 21000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 5.88,
|
| 558 |
+
"learning_rate": 6e-05,
|
| 559 |
+
"loss": 3.9309,
|
| 560 |
+
"step": 21500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 5.88,
|
| 564 |
+
"eval_loss": 3.7906742095947266,
|
| 565 |
+
"eval_runtime": 9.4675,
|
| 566 |
+
"eval_samples_per_second": 24.927,
|
| 567 |
+
"step": 21500
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 6.02,
|
| 571 |
+
"learning_rate": 6e-05,
|
| 572 |
+
"loss": 3.9105,
|
| 573 |
+
"step": 22000
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 6.02,
|
| 577 |
+
"eval_loss": 3.7875397205352783,
|
| 578 |
+
"eval_runtime": 9.4606,
|
| 579 |
+
"eval_samples_per_second": 24.946,
|
| 580 |
+
"step": 22000
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 6.16,
|
| 584 |
+
"learning_rate": 6e-05,
|
| 585 |
+
"loss": 3.8424,
|
| 586 |
+
"step": 22500
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 6.16,
|
| 590 |
+
"eval_loss": 3.782013416290283,
|
| 591 |
+
"eval_runtime": 9.4639,
|
| 592 |
+
"eval_samples_per_second": 24.937,
|
| 593 |
+
"step": 22500
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 6.29,
|
| 597 |
+
"learning_rate": 6e-05,
|
| 598 |
+
"loss": 3.8579,
|
| 599 |
+
"step": 23000
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 6.29,
|
| 603 |
+
"eval_loss": 3.781845808029175,
|
| 604 |
+
"eval_runtime": 9.4621,
|
| 605 |
+
"eval_samples_per_second": 24.941,
|
| 606 |
+
"step": 23000
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 6.43,
|
| 610 |
+
"learning_rate": 6e-05,
|
| 611 |
+
"loss": 3.868,
|
| 612 |
+
"step": 23500
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 6.43,
|
| 616 |
+
"eval_loss": 3.76729679107666,
|
| 617 |
+
"eval_runtime": 9.4657,
|
| 618 |
+
"eval_samples_per_second": 24.932,
|
| 619 |
+
"step": 23500
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 6.57,
|
| 623 |
+
"learning_rate": 6e-05,
|
| 624 |
+
"loss": 3.8631,
|
| 625 |
+
"step": 24000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 6.57,
|
| 629 |
+
"eval_loss": 3.7602250576019287,
|
| 630 |
+
"eval_runtime": 9.4649,
|
| 631 |
+
"eval_samples_per_second": 24.934,
|
| 632 |
+
"step": 24000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 6.7,
|
| 636 |
+
"learning_rate": 6e-05,
|
| 637 |
+
"loss": 3.8635,
|
| 638 |
+
"step": 24500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 6.7,
|
| 642 |
+
"eval_loss": 3.7623238563537598,
|
| 643 |
+
"eval_runtime": 9.456,
|
| 644 |
+
"eval_samples_per_second": 24.958,
|
| 645 |
+
"step": 24500
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 6.84,
|
| 649 |
+
"learning_rate": 6e-05,
|
| 650 |
+
"loss": 3.8632,
|
| 651 |
+
"step": 25000
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 6.84,
|
| 655 |
+
"eval_loss": 3.7607743740081787,
|
| 656 |
+
"eval_runtime": 9.4601,
|
| 657 |
+
"eval_samples_per_second": 24.947,
|
| 658 |
+
"step": 25000
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 6.98,
|
| 662 |
+
"learning_rate": 6e-05,
|
| 663 |
+
"loss": 3.873,
|
| 664 |
+
"step": 25500
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 6.98,
|
| 668 |
+
"eval_loss": 3.749258279800415,
|
| 669 |
+
"eval_runtime": 9.4598,
|
| 670 |
+
"eval_samples_per_second": 24.948,
|
| 671 |
+
"step": 25500
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 7.11,
|
| 675 |
+
"learning_rate": 6e-05,
|
| 676 |
+
"loss": 3.7911,
|
| 677 |
+
"step": 26000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 7.11,
|
| 681 |
+
"eval_loss": 3.744161367416382,
|
| 682 |
+
"eval_runtime": 9.4582,
|
| 683 |
+
"eval_samples_per_second": 24.952,
|
| 684 |
+
"step": 26000
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 7.25,
|
| 688 |
+
"learning_rate": 6e-05,
|
| 689 |
+
"loss": 3.8039,
|
| 690 |
+
"step": 26500
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 7.25,
|
| 694 |
+
"eval_loss": 3.739082098007202,
|
| 695 |
+
"eval_runtime": 9.4581,
|
| 696 |
+
"eval_samples_per_second": 24.952,
|
| 697 |
+
"step": 26500
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 7.39,
|
| 701 |
+
"learning_rate": 6e-05,
|
| 702 |
+
"loss": 3.7996,
|
| 703 |
+
"step": 27000
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 7.39,
|
| 707 |
+
"eval_loss": 3.738431692123413,
|
| 708 |
+
"eval_runtime": 9.4714,
|
| 709 |
+
"eval_samples_per_second": 24.917,
|
| 710 |
+
"step": 27000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 7.52,
|
| 714 |
+
"learning_rate": 6e-05,
|
| 715 |
+
"loss": 3.8157,
|
| 716 |
+
"step": 27500
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 7.52,
|
| 720 |
+
"eval_loss": 3.7242279052734375,
|
| 721 |
+
"eval_runtime": 9.4626,
|
| 722 |
+
"eval_samples_per_second": 24.94,
|
| 723 |
+
"step": 27500
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 7.66,
|
| 727 |
+
"learning_rate": 6e-05,
|
| 728 |
+
"loss": 3.8224,
|
| 729 |
+
"step": 28000
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 7.66,
|
| 733 |
+
"eval_loss": 3.7267138957977295,
|
| 734 |
+
"eval_runtime": 9.4527,
|
| 735 |
+
"eval_samples_per_second": 24.966,
|
| 736 |
+
"step": 28000
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 7.8,
|
| 740 |
+
"learning_rate": 6e-05,
|
| 741 |
+
"loss": 3.8246,
|
| 742 |
+
"step": 28500
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 7.8,
|
| 746 |
+
"eval_loss": 3.711819887161255,
|
| 747 |
+
"eval_runtime": 9.4622,
|
| 748 |
+
"eval_samples_per_second": 24.941,
|
| 749 |
+
"step": 28500
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 7.93,
|
| 753 |
+
"learning_rate": 6e-05,
|
| 754 |
+
"loss": 3.8176,
|
| 755 |
+
"step": 29000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 7.93,
|
| 759 |
+
"eval_loss": 3.707026958465576,
|
| 760 |
+
"eval_runtime": 9.4595,
|
| 761 |
+
"eval_samples_per_second": 24.948,
|
| 762 |
+
"step": 29000
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 8.07,
|
| 766 |
+
"learning_rate": 6e-05,
|
| 767 |
+
"loss": 3.7763,
|
| 768 |
+
"step": 29500
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 8.07,
|
| 772 |
+
"eval_loss": 3.7134788036346436,
|
| 773 |
+
"eval_runtime": 9.4603,
|
| 774 |
+
"eval_samples_per_second": 24.946,
|
| 775 |
+
"step": 29500
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 8.21,
|
| 779 |
+
"learning_rate": 6e-05,
|
| 780 |
+
"loss": 3.7557,
|
| 781 |
+
"step": 30000
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 8.21,
|
| 785 |
+
"eval_loss": 3.7067065238952637,
|
| 786 |
+
"eval_runtime": 9.4683,
|
| 787 |
+
"eval_samples_per_second": 24.925,
|
| 788 |
+
"step": 30000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 8.34,
|
| 792 |
+
"learning_rate": 6e-05,
|
| 793 |
+
"loss": 3.7662,
|
| 794 |
+
"step": 30500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 8.34,
|
| 798 |
+
"eval_loss": 3.7023823261260986,
|
| 799 |
+
"eval_runtime": 9.4321,
|
| 800 |
+
"eval_samples_per_second": 25.021,
|
| 801 |
+
"step": 30500
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 8.48,
|
| 805 |
+
"learning_rate": 6e-05,
|
| 806 |
+
"loss": 3.7677,
|
| 807 |
+
"step": 31000
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 8.48,
|
| 811 |
+
"eval_loss": 3.699657678604126,
|
| 812 |
+
"eval_runtime": 9.4485,
|
| 813 |
+
"eval_samples_per_second": 24.978,
|
| 814 |
+
"step": 31000
|
| 815 |
+
}
|
| 816 |
+
],
|
| 817 |
+
"max_steps": 36550,
|
| 818 |
+
"num_train_epochs": 10,
|
| 819 |
+
"total_flos": 3969647640576000.0,
|
| 820 |
+
"trial_name": null,
|
| 821 |
+
"trial_params": null
|
| 822 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1bdcb04ef96c10bb63e37c24fd535e0062cd403e2f184623b1708e684f818c9
|
| 3 |
+
size 2479
|