Commit
•
7f721c2
1
Parent(s):
f3b6db4
Model save
Browse files- README.md +85 -0
- adapter_model.safetensors +1 -1
- all_results.json +21 -0
- eval_results.json +16 -0
- runs/May10_17-48-57_gpu4-119-5/events.out.tfevents.1715327400.gpu4-119-5.3130064.0 +2 -2
- runs/May10_17-48-57_gpu4-119-5/events.out.tfevents.1715350020.gpu4-119-5.3130064.1 +3 -0
- train_results.json +8 -0
- trainer_state.json +2014 -0
README.md
ADDED
|
@@ -0,0 +1,85 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
tags:
|
| 4 |
+
- trl
|
| 5 |
+
- dpo
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
base_model: DUAL-GPO/phi-2-sft-lora-ultrachat-merged
|
| 8 |
+
model-index:
|
| 9 |
+
- name: phi-2-gpo-newSFT-b0.001-i0
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# phi-2-gpo-newSFT-b0.001-i0
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [DUAL-GPO/phi-2-sft-lora-ultrachat-merged](https://huggingface.co/DUAL-GPO/phi-2-sft-lora-ultrachat-merged) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.0401
|
| 21 |
+
- Rewards/chosen: -0.0556
|
| 22 |
+
- Rewards/rejected: -0.0895
|
| 23 |
+
- Rewards/accuracies: 0.6018
|
| 24 |
+
- Rewards/margins: 0.0338
|
| 25 |
+
- Logps/rejected: -337.8973
|
| 26 |
+
- Logps/chosen: -325.6151
|
| 27 |
+
- Logits/rejected: 0.2226
|
| 28 |
+
- Logits/chosen: 0.1916
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 5e-06
|
| 48 |
+
- train_batch_size: 4
|
| 49 |
+
- eval_batch_size: 4
|
| 50 |
+
- seed: 42
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 3
|
| 53 |
+
- gradient_accumulation_steps: 4
|
| 54 |
+
- total_train_batch_size: 48
|
| 55 |
+
- total_eval_batch_size: 12
|
| 56 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
+
- lr_scheduler_type: cosine
|
| 58 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 59 |
+
- num_epochs: 1
|
| 60 |
+
|
| 61 |
+
### Training results
|
| 62 |
+
|
| 63 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 64 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 65 |
+
| 0.0569 | 0.08 | 100 | 0.0535 | 0.0003 | -0.0004 | 0.5973 | 0.0006 | -248.8037 | -269.7070 | 1.0547 | 0.9877 |
|
| 66 |
+
| 0.0565 | 0.16 | 200 | 0.0515 | -0.0016 | -0.0068 | 0.6078 | 0.0052 | -255.2234 | -271.5318 | 0.9706 | 0.8998 |
|
| 67 |
+
| 0.0467 | 0.24 | 300 | 0.0470 | -0.0316 | -0.0485 | 0.6033 | 0.0169 | -296.9762 | -301.5743 | 0.5528 | 0.4973 |
|
| 68 |
+
| 0.0467 | 0.31 | 400 | 0.0443 | -0.0370 | -0.0583 | 0.6033 | 0.0213 | -306.7241 | -306.9802 | 0.3457 | 0.3071 |
|
| 69 |
+
| 0.0359 | 0.39 | 500 | 0.0428 | -0.0574 | -0.0869 | 0.5988 | 0.0296 | -335.3609 | -327.3275 | 0.2119 | 0.1835 |
|
| 70 |
+
| 0.0431 | 0.47 | 600 | 0.0418 | -0.0450 | -0.0725 | 0.6033 | 0.0275 | -320.9161 | -314.9630 | 0.2891 | 0.2554 |
|
| 71 |
+
| 0.0438 | 0.55 | 700 | 0.0413 | -0.0574 | -0.0889 | 0.6018 | 0.0316 | -337.3519 | -327.3254 | 0.2356 | 0.2040 |
|
| 72 |
+
| 0.0446 | 0.63 | 800 | 0.0409 | -0.0522 | -0.0842 | 0.6048 | 0.0320 | -332.6603 | -322.1777 | 0.2566 | 0.2236 |
|
| 73 |
+
| 0.0426 | 0.71 | 900 | 0.0408 | -0.0624 | -0.0977 | 0.6048 | 0.0353 | -346.1494 | -332.3424 | 0.2089 | 0.1797 |
|
| 74 |
+
| 0.0448 | 0.79 | 1000 | 0.0403 | -0.0545 | -0.0869 | 0.6063 | 0.0324 | -335.3596 | -324.4480 | 0.2463 | 0.2141 |
|
| 75 |
+
| 0.0411 | 0.86 | 1100 | 0.0402 | -0.0549 | -0.0884 | 0.6018 | 0.0336 | -336.8657 | -324.8257 | 0.2283 | 0.1971 |
|
| 76 |
+
| 0.0459 | 0.94 | 1200 | 0.0401 | -0.0558 | -0.0896 | 0.6033 | 0.0338 | -338.0042 | -325.7257 | 0.2205 | 0.1898 |
|
| 77 |
+
|
| 78 |
+
|
| 79 |
+
### Framework versions
|
| 80 |
+
|
| 81 |
+
- PEFT 0.7.1
|
| 82 |
+
- Transformers 4.36.2
|
| 83 |
+
- Pytorch 2.1.2
|
| 84 |
+
- Datasets 2.14.6
|
| 85 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 167807296
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5d6a4b684b3480736d65d956434f794dceef05509c17c5044baaf501f579698
|
| 3 |
size 167807296
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": 0.1915939450263977,
|
| 4 |
+
"eval_logits/rejected": 0.22264356911182404,
|
| 5 |
+
"eval_logps/chosen": -325.6150817871094,
|
| 6 |
+
"eval_logps/rejected": -337.89727783203125,
|
| 7 |
+
"eval_loss": 0.04012390971183777,
|
| 8 |
+
"eval_rewards/accuracies": 0.6017963886260986,
|
| 9 |
+
"eval_rewards/chosen": -0.05564659833908081,
|
| 10 |
+
"eval_rewards/margins": 0.033811911940574646,
|
| 11 |
+
"eval_rewards/rejected": -0.08945851027965546,
|
| 12 |
+
"eval_runtime": 374.9506,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 5.334,
|
| 15 |
+
"eval_steps_per_second": 0.445,
|
| 16 |
+
"train_loss": 0.044965725131654775,
|
| 17 |
+
"train_runtime": 22244.9719,
|
| 18 |
+
"train_samples": 61135,
|
| 19 |
+
"train_samples_per_second": 2.748,
|
| 20 |
+
"train_steps_per_second": 0.057
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_logits/chosen": 0.1915939450263977,
|
| 4 |
+
"eval_logits/rejected": 0.22264356911182404,
|
| 5 |
+
"eval_logps/chosen": -325.6150817871094,
|
| 6 |
+
"eval_logps/rejected": -337.89727783203125,
|
| 7 |
+
"eval_loss": 0.04012390971183777,
|
| 8 |
+
"eval_rewards/accuracies": 0.6017963886260986,
|
| 9 |
+
"eval_rewards/chosen": -0.05564659833908081,
|
| 10 |
+
"eval_rewards/margins": 0.033811911940574646,
|
| 11 |
+
"eval_rewards/rejected": -0.08945851027965546,
|
| 12 |
+
"eval_runtime": 374.9506,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 5.334,
|
| 15 |
+
"eval_steps_per_second": 0.445
|
| 16 |
+
}
|
runs/May10_17-48-57_gpu4-119-5/events.out.tfevents.1715327400.gpu4-119-5.3130064.0
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:c0acb8689ff2cef83dba6d4c174fec09536e06b56ac84fa635d32092a7cb1ec0
|
| 3 |
+
size 95143
|
runs/May10_17-48-57_gpu4-119-5/events.out.tfevents.1715350020.gpu4-119-5.3130064.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b21c58b6047a2e747f19428d6630c19e56d4607eb68e8d26a7f3b47f8f24a19
|
| 3 |
+
size 828
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.044965725131654775,
|
| 4 |
+
"train_runtime": 22244.9719,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 2.748,
|
| 7 |
+
"train_steps_per_second": 0.057
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2014 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9994111874386653,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1273,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 3.90625e-08,
|
| 14 |
+
"logits/chosen": 0.8174187541007996,
|
| 15 |
+
"logits/rejected": 1.0011495351791382,
|
| 16 |
+
"logps/chosen": -246.74014282226562,
|
| 17 |
+
"logps/rejected": -187.51809692382812,
|
| 18 |
+
"loss": 0.0442,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.01,
|
| 27 |
+
"learning_rate": 3.90625e-07,
|
| 28 |
+
"logits/chosen": 1.018452525138855,
|
| 29 |
+
"logits/rejected": 1.0691548585891724,
|
| 30 |
+
"logps/chosen": -264.271240234375,
|
| 31 |
+
"logps/rejected": -221.92071533203125,
|
| 32 |
+
"loss": 0.0614,
|
| 33 |
+
"rewards/accuracies": 0.4444444477558136,
|
| 34 |
+
"rewards/chosen": 1.4848755199636798e-05,
|
| 35 |
+
"rewards/margins": 9.348281309939921e-05,
|
| 36 |
+
"rewards/rejected": -7.86340533522889e-05,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02,
|
| 41 |
+
"learning_rate": 7.8125e-07,
|
| 42 |
+
"logits/chosen": 0.962690532207489,
|
| 43 |
+
"logits/rejected": 0.9728161096572876,
|
| 44 |
+
"logps/chosen": -255.00430297851562,
|
| 45 |
+
"logps/rejected": -221.725830078125,
|
| 46 |
+
"loss": 0.0504,
|
| 47 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 48 |
+
"rewards/chosen": -1.0949605893983971e-05,
|
| 49 |
+
"rewards/margins": 4.187399099464528e-05,
|
| 50 |
+
"rewards/rejected": -5.282359052216634e-05,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.02,
|
| 55 |
+
"learning_rate": 1.1718750000000001e-06,
|
| 56 |
+
"logits/chosen": 1.0629122257232666,
|
| 57 |
+
"logits/rejected": 1.116347074508667,
|
| 58 |
+
"logps/chosen": -241.6275177001953,
|
| 59 |
+
"logps/rejected": -226.95309448242188,
|
| 60 |
+
"loss": 0.0613,
|
| 61 |
+
"rewards/accuracies": 0.41874998807907104,
|
| 62 |
+
"rewards/chosen": -4.63350661448203e-05,
|
| 63 |
+
"rewards/margins": -5.460592728923075e-05,
|
| 64 |
+
"rewards/rejected": 8.270866601378657e-06,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.03,
|
| 69 |
+
"learning_rate": 1.5625e-06,
|
| 70 |
+
"logits/chosen": 0.9653002023696899,
|
| 71 |
+
"logits/rejected": 1.0042797327041626,
|
| 72 |
+
"logps/chosen": -252.0692596435547,
|
| 73 |
+
"logps/rejected": -237.03872680664062,
|
| 74 |
+
"loss": 0.0455,
|
| 75 |
+
"rewards/accuracies": 0.512499988079071,
|
| 76 |
+
"rewards/chosen": 9.635712558520027e-06,
|
| 77 |
+
"rewards/margins": 7.444314542226493e-05,
|
| 78 |
+
"rewards/rejected": -6.480743468273431e-05,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.04,
|
| 83 |
+
"learning_rate": 1.953125e-06,
|
| 84 |
+
"logits/chosen": 0.9830142259597778,
|
| 85 |
+
"logits/rejected": 1.056870460510254,
|
| 86 |
+
"logps/chosen": -256.4426574707031,
|
| 87 |
+
"logps/rejected": -225.88821411132812,
|
| 88 |
+
"loss": 0.0536,
|
| 89 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 90 |
+
"rewards/chosen": -2.2923812139197253e-05,
|
| 91 |
+
"rewards/margins": 0.00013652857160195708,
|
| 92 |
+
"rewards/rejected": -0.00015945239283610135,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.05,
|
| 97 |
+
"learning_rate": 2.3437500000000002e-06,
|
| 98 |
+
"logits/chosen": 0.9784016609191895,
|
| 99 |
+
"logits/rejected": 1.1101980209350586,
|
| 100 |
+
"logps/chosen": -271.16668701171875,
|
| 101 |
+
"logps/rejected": -249.8975372314453,
|
| 102 |
+
"loss": 0.0572,
|
| 103 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 104 |
+
"rewards/chosen": 2.5281901798734907e-06,
|
| 105 |
+
"rewards/margins": 0.00010906215902650729,
|
| 106 |
+
"rewards/rejected": -0.00010653395293047652,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.05,
|
| 111 |
+
"learning_rate": 2.7343750000000004e-06,
|
| 112 |
+
"logits/chosen": 1.0381357669830322,
|
| 113 |
+
"logits/rejected": 1.0707155466079712,
|
| 114 |
+
"logps/chosen": -253.42160034179688,
|
| 115 |
+
"logps/rejected": -221.3090362548828,
|
| 116 |
+
"loss": 0.0465,
|
| 117 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 118 |
+
"rewards/chosen": 9.924682672135532e-06,
|
| 119 |
+
"rewards/margins": 0.00015668092237319797,
|
| 120 |
+
"rewards/rejected": -0.0001467562251491472,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.06,
|
| 125 |
+
"learning_rate": 3.125e-06,
|
| 126 |
+
"logits/chosen": 0.9301819801330566,
|
| 127 |
+
"logits/rejected": 1.1145567893981934,
|
| 128 |
+
"logps/chosen": -291.00421142578125,
|
| 129 |
+
"logps/rejected": -223.3363800048828,
|
| 130 |
+
"loss": 0.0539,
|
| 131 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 132 |
+
"rewards/chosen": 6.633991142734885e-05,
|
| 133 |
+
"rewards/margins": 0.00031302389106713235,
|
| 134 |
+
"rewards/rejected": -0.0002466839796397835,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.07,
|
| 139 |
+
"learning_rate": 3.5156250000000003e-06,
|
| 140 |
+
"logits/chosen": 1.054889440536499,
|
| 141 |
+
"logits/rejected": 1.0736507177352905,
|
| 142 |
+
"logps/chosen": -287.6363830566406,
|
| 143 |
+
"logps/rejected": -294.4855651855469,
|
| 144 |
+
"loss": 0.0478,
|
| 145 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 146 |
+
"rewards/chosen": 0.0002254299761261791,
|
| 147 |
+
"rewards/margins": 0.0004840657056774944,
|
| 148 |
+
"rewards/rejected": -0.0002586357877589762,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.08,
|
| 153 |
+
"learning_rate": 3.90625e-06,
|
| 154 |
+
"logits/chosen": 0.9826571345329285,
|
| 155 |
+
"logits/rejected": 1.0136107206344604,
|
| 156 |
+
"logps/chosen": -248.9982147216797,
|
| 157 |
+
"logps/rejected": -213.6271514892578,
|
| 158 |
+
"loss": 0.0569,
|
| 159 |
+
"rewards/accuracies": 0.612500011920929,
|
| 160 |
+
"rewards/chosen": 0.00023196591064333916,
|
| 161 |
+
"rewards/margins": 0.0005559330456890166,
|
| 162 |
+
"rewards/rejected": -0.0003239671350456774,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.08,
|
| 167 |
+
"eval_logits/chosen": 0.987690269947052,
|
| 168 |
+
"eval_logits/rejected": 1.0547419786453247,
|
| 169 |
+
"eval_logps/chosen": -269.70703125,
|
| 170 |
+
"eval_logps/rejected": -248.80368041992188,
|
| 171 |
+
"eval_loss": 0.053490906953811646,
|
| 172 |
+
"eval_rewards/accuracies": 0.597305417060852,
|
| 173 |
+
"eval_rewards/chosen": 0.0002614832192193717,
|
| 174 |
+
"eval_rewards/margins": 0.0006264070980250835,
|
| 175 |
+
"eval_rewards/rejected": -0.00036492382059805095,
|
| 176 |
+
"eval_runtime": 374.224,
|
| 177 |
+
"eval_samples_per_second": 5.344,
|
| 178 |
+
"eval_steps_per_second": 0.446,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.09,
|
| 183 |
+
"learning_rate": 4.296875e-06,
|
| 184 |
+
"logits/chosen": 1.0466687679290771,
|
| 185 |
+
"logits/rejected": 0.9887654185295105,
|
| 186 |
+
"logps/chosen": -299.6304931640625,
|
| 187 |
+
"logps/rejected": -244.9992218017578,
|
| 188 |
+
"loss": 0.0627,
|
| 189 |
+
"rewards/accuracies": 0.59375,
|
| 190 |
+
"rewards/chosen": 0.0005523829022422433,
|
| 191 |
+
"rewards/margins": 0.0007637250819243491,
|
| 192 |
+
"rewards/rejected": -0.0002113422378897667,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.09,
|
| 197 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 198 |
+
"logits/chosen": 1.0038639307022095,
|
| 199 |
+
"logits/rejected": 1.0830211639404297,
|
| 200 |
+
"logps/chosen": -273.029296875,
|
| 201 |
+
"logps/rejected": -203.0375518798828,
|
| 202 |
+
"loss": 0.0646,
|
| 203 |
+
"rewards/accuracies": 0.574999988079071,
|
| 204 |
+
"rewards/chosen": 0.0005045495927333832,
|
| 205 |
+
"rewards/margins": 0.0008580017019994557,
|
| 206 |
+
"rewards/rejected": -0.0003534521092660725,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.1,
|
| 211 |
+
"learning_rate": 4.999962359300416e-06,
|
| 212 |
+
"logits/chosen": 1.0437225103378296,
|
| 213 |
+
"logits/rejected": 1.1165318489074707,
|
| 214 |
+
"logps/chosen": -275.69549560546875,
|
| 215 |
+
"logps/rejected": -231.8827667236328,
|
| 216 |
+
"loss": 0.0515,
|
| 217 |
+
"rewards/accuracies": 0.581250011920929,
|
| 218 |
+
"rewards/chosen": 0.0014680480817332864,
|
| 219 |
+
"rewards/margins": 0.0015463808085769415,
|
| 220 |
+
"rewards/rejected": -7.833284325897694e-05,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.11,
|
| 225 |
+
"learning_rate": 4.998645053824218e-06,
|
| 226 |
+
"logits/chosen": 1.012627363204956,
|
| 227 |
+
"logits/rejected": 1.0429273843765259,
|
| 228 |
+
"logps/chosen": -306.0951232910156,
|
| 229 |
+
"logps/rejected": -257.19134521484375,
|
| 230 |
+
"loss": 0.052,
|
| 231 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 232 |
+
"rewards/chosen": 0.0023107831366360188,
|
| 233 |
+
"rewards/margins": 0.0023487000726163387,
|
| 234 |
+
"rewards/rejected": -3.7916899600531906e-05,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.12,
|
| 239 |
+
"learning_rate": 4.9954468466732145e-06,
|
| 240 |
+
"logits/chosen": 0.9628115892410278,
|
| 241 |
+
"logits/rejected": 1.01333749294281,
|
| 242 |
+
"logps/chosen": -274.50750732421875,
|
| 243 |
+
"logps/rejected": -251.54348754882812,
|
| 244 |
+
"loss": 0.0499,
|
| 245 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 246 |
+
"rewards/chosen": 0.0016668376047164202,
|
| 247 |
+
"rewards/margins": 0.001496818964369595,
|
| 248 |
+
"rewards/rejected": 0.00017001861124299467,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.13,
|
| 253 |
+
"learning_rate": 4.990370145357496e-06,
|
| 254 |
+
"logits/chosen": 0.9540281295776367,
|
| 255 |
+
"logits/rejected": 1.0164401531219482,
|
| 256 |
+
"logps/chosen": -245.35122680664062,
|
| 257 |
+
"logps/rejected": -222.86331176757812,
|
| 258 |
+
"loss": 0.0556,
|
| 259 |
+
"rewards/accuracies": 0.59375,
|
| 260 |
+
"rewards/chosen": 0.0014593935338780284,
|
| 261 |
+
"rewards/margins": 0.0026508464943617582,
|
| 262 |
+
"rewards/rejected": -0.001191452844068408,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.13,
|
| 267 |
+
"learning_rate": 4.983418771458684e-06,
|
| 268 |
+
"logits/chosen": 1.0378397703170776,
|
| 269 |
+
"logits/rejected": 1.0483719110488892,
|
| 270 |
+
"logps/chosen": -299.852294921875,
|
| 271 |
+
"logps/rejected": -257.53851318359375,
|
| 272 |
+
"loss": 0.0533,
|
| 273 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 274 |
+
"rewards/chosen": 0.002016807906329632,
|
| 275 |
+
"rewards/margins": 0.002726987935602665,
|
| 276 |
+
"rewards/rejected": -0.000710180145688355,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.14,
|
| 281 |
+
"learning_rate": 4.97459795775315e-06,
|
| 282 |
+
"logits/chosen": 0.9879986047744751,
|
| 283 |
+
"logits/rejected": 1.0183862447738647,
|
| 284 |
+
"logps/chosen": -292.57647705078125,
|
| 285 |
+
"logps/rejected": -274.23297119140625,
|
| 286 |
+
"loss": 0.0545,
|
| 287 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 288 |
+
"rewards/chosen": 0.003255961462855339,
|
| 289 |
+
"rewards/margins": 0.004640574567019939,
|
| 290 |
+
"rewards/rejected": -0.0013846131041646004,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.15,
|
| 295 |
+
"learning_rate": 4.963914344272961e-06,
|
| 296 |
+
"logits/chosen": 0.928981602191925,
|
| 297 |
+
"logits/rejected": 1.0115787982940674,
|
| 298 |
+
"logps/chosen": -274.09576416015625,
|
| 299 |
+
"logps/rejected": -235.8800506591797,
|
| 300 |
+
"loss": 0.0498,
|
| 301 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 302 |
+
"rewards/chosen": 0.0013791031669825315,
|
| 303 |
+
"rewards/margins": 0.004472037777304649,
|
| 304 |
+
"rewards/rejected": -0.0030929348431527615,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.16,
|
| 309 |
+
"learning_rate": 4.951375973307458e-06,
|
| 310 |
+
"logits/chosen": 0.9295806884765625,
|
| 311 |
+
"logits/rejected": 0.9614816904067993,
|
| 312 |
+
"logps/chosen": -267.1017761230469,
|
| 313 |
+
"logps/rejected": -243.70095825195312,
|
| 314 |
+
"loss": 0.0565,
|
| 315 |
+
"rewards/accuracies": 0.581250011920929,
|
| 316 |
+
"rewards/chosen": -0.0006091211689636111,
|
| 317 |
+
"rewards/margins": 0.0036107245832681656,
|
| 318 |
+
"rewards/rejected": -0.004219844937324524,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.16,
|
| 323 |
+
"eval_logits/chosen": 0.8997535705566406,
|
| 324 |
+
"eval_logits/rejected": 0.9705783128738403,
|
| 325 |
+
"eval_logps/chosen": -271.5317687988281,
|
| 326 |
+
"eval_logps/rejected": -255.22335815429688,
|
| 327 |
+
"eval_loss": 0.051502373069524765,
|
| 328 |
+
"eval_rewards/accuracies": 0.6077844500541687,
|
| 329 |
+
"eval_rewards/chosen": -0.001563269179314375,
|
| 330 |
+
"eval_rewards/margins": 0.005221350584179163,
|
| 331 |
+
"eval_rewards/rejected": -0.006784618832170963,
|
| 332 |
+
"eval_runtime": 374.3563,
|
| 333 |
+
"eval_samples_per_second": 5.343,
|
| 334 |
+
"eval_steps_per_second": 0.446,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.16,
|
| 339 |
+
"learning_rate": 4.93699228334928e-06,
|
| 340 |
+
"logits/chosen": 0.9301688075065613,
|
| 341 |
+
"logits/rejected": 0.9155396223068237,
|
| 342 |
+
"logps/chosen": -285.2063293457031,
|
| 343 |
+
"logps/rejected": -238.8046417236328,
|
| 344 |
+
"loss": 0.0507,
|
| 345 |
+
"rewards/accuracies": 0.643750011920929,
|
| 346 |
+
"rewards/chosen": -0.00031993992161005735,
|
| 347 |
+
"rewards/margins": 0.006842759437859058,
|
| 348 |
+
"rewards/rejected": -0.00716269901022315,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.17,
|
| 353 |
+
"learning_rate": 4.920774101989362e-06,
|
| 354 |
+
"logits/chosen": 0.9102910757064819,
|
| 355 |
+
"logits/rejected": 0.9553702473640442,
|
| 356 |
+
"logps/chosen": -285.9295959472656,
|
| 357 |
+
"logps/rejected": -256.9110107421875,
|
| 358 |
+
"loss": 0.0525,
|
| 359 |
+
"rewards/accuracies": 0.612500011920929,
|
| 360 |
+
"rewards/chosen": -0.0035760626196861267,
|
| 361 |
+
"rewards/margins": 0.005626793950796127,
|
| 362 |
+
"rewards/rejected": -0.00920285563915968,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.18,
|
| 367 |
+
"learning_rate": 4.902733637766261e-06,
|
| 368 |
+
"logits/chosen": 0.8825628161430359,
|
| 369 |
+
"logits/rejected": 0.9233707189559937,
|
| 370 |
+
"logps/chosen": -299.63818359375,
|
| 371 |
+
"logps/rejected": -264.28070068359375,
|
| 372 |
+
"loss": 0.0542,
|
| 373 |
+
"rewards/accuracies": 0.625,
|
| 374 |
+
"rewards/chosen": -0.004372667986899614,
|
| 375 |
+
"rewards/margins": 0.010200297459959984,
|
| 376 |
+
"rewards/rejected": -0.014572965912520885,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.19,
|
| 381 |
+
"learning_rate": 4.882884470975954e-06,
|
| 382 |
+
"logits/chosen": 0.8268105387687683,
|
| 383 |
+
"logits/rejected": 0.9335943460464478,
|
| 384 |
+
"logps/chosen": -257.7367248535156,
|
| 385 |
+
"logps/rejected": -233.0779571533203,
|
| 386 |
+
"loss": 0.0534,
|
| 387 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 388 |
+
"rewards/chosen": -0.010581925511360168,
|
| 389 |
+
"rewards/margins": 0.006359722465276718,
|
| 390 |
+
"rewards/rejected": -0.016941647976636887,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.2,
|
| 395 |
+
"learning_rate": 4.861241543449015e-06,
|
| 396 |
+
"logits/chosen": 0.8305758237838745,
|
| 397 |
+
"logits/rejected": 0.833528995513916,
|
| 398 |
+
"logps/chosen": -299.9106140136719,
|
| 399 |
+
"logps/rejected": -286.69024658203125,
|
| 400 |
+
"loss": 0.0455,
|
| 401 |
+
"rewards/accuracies": 0.59375,
|
| 402 |
+
"rewards/chosen": -0.014483618550002575,
|
| 403 |
+
"rewards/margins": 0.007877354510128498,
|
| 404 |
+
"rewards/rejected": -0.022360973060131073,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.2,
|
| 409 |
+
"learning_rate": 4.8378211473028755e-06,
|
| 410 |
+
"logits/chosen": 0.793732762336731,
|
| 411 |
+
"logits/rejected": 0.7972532510757446,
|
| 412 |
+
"logps/chosen": -286.5875244140625,
|
| 413 |
+
"logps/rejected": -260.48895263671875,
|
| 414 |
+
"loss": 0.048,
|
| 415 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 416 |
+
"rewards/chosen": -0.01467740349471569,
|
| 417 |
+
"rewards/margins": 0.01562904380261898,
|
| 418 |
+
"rewards/rejected": -0.03030644915997982,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.21,
|
| 423 |
+
"learning_rate": 4.812640912677624e-06,
|
| 424 |
+
"logits/chosen": 0.6651993989944458,
|
| 425 |
+
"logits/rejected": 0.7179329991340637,
|
| 426 |
+
"logps/chosen": -272.69989013671875,
|
| 427 |
+
"logps/rejected": -246.05810546875,
|
| 428 |
+
"loss": 0.0557,
|
| 429 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 430 |
+
"rewards/chosen": -0.021714814007282257,
|
| 431 |
+
"rewards/margins": 0.006616706494241953,
|
| 432 |
+
"rewards/rejected": -0.028331521898508072,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.22,
|
| 437 |
+
"learning_rate": 4.785719794464596e-06,
|
| 438 |
+
"logits/chosen": 0.5646941065788269,
|
| 439 |
+
"logits/rejected": 0.6421071290969849,
|
| 440 |
+
"logps/chosen": -334.10302734375,
|
| 441 |
+
"logps/rejected": -274.1715087890625,
|
| 442 |
+
"loss": 0.046,
|
| 443 |
+
"rewards/accuracies": 0.625,
|
| 444 |
+
"rewards/chosen": -0.023918839171528816,
|
| 445 |
+
"rewards/margins": 0.012724781408905983,
|
| 446 |
+
"rewards/rejected": -0.0366436168551445,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.23,
|
| 451 |
+
"learning_rate": 4.757078058037722e-06,
|
| 452 |
+
"logits/chosen": 0.5190677046775818,
|
| 453 |
+
"logits/rejected": 0.6334066987037659,
|
| 454 |
+
"logps/chosen": -340.30914306640625,
|
| 455 |
+
"logps/rejected": -305.35662841796875,
|
| 456 |
+
"loss": 0.0475,
|
| 457 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 458 |
+
"rewards/chosen": -0.021180950105190277,
|
| 459 |
+
"rewards/margins": 0.02034229040145874,
|
| 460 |
+
"rewards/rejected": -0.04152324050664902,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.24,
|
| 465 |
+
"learning_rate": 4.72673726399839e-06,
|
| 466 |
+
"logits/chosen": 0.5193800330162048,
|
| 467 |
+
"logits/rejected": 0.6493593454360962,
|
| 468 |
+
"logps/chosen": -261.5180358886719,
|
| 469 |
+
"logps/rejected": -240.78369140625,
|
| 470 |
+
"loss": 0.0467,
|
| 471 |
+
"rewards/accuracies": 0.5625,
|
| 472 |
+
"rewards/chosen": -0.032134272158145905,
|
| 473 |
+
"rewards/margins": 0.01347540132701397,
|
| 474 |
+
"rewards/rejected": -0.045609671622514725,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.24,
|
| 479 |
+
"eval_logits/chosen": 0.4972836971282959,
|
| 480 |
+
"eval_logits/rejected": 0.5527552962303162,
|
| 481 |
+
"eval_logps/chosen": -301.5743103027344,
|
| 482 |
+
"eval_logps/rejected": -296.9762268066406,
|
| 483 |
+
"eval_loss": 0.04698715731501579,
|
| 484 |
+
"eval_rewards/accuracies": 0.6032934188842773,
|
| 485 |
+
"eval_rewards/chosen": -0.0316057950258255,
|
| 486 |
+
"eval_rewards/margins": 0.016931703314185143,
|
| 487 |
+
"eval_rewards/rejected": -0.04853750020265579,
|
| 488 |
+
"eval_runtime": 374.353,
|
| 489 |
+
"eval_samples_per_second": 5.343,
|
| 490 |
+
"eval_steps_per_second": 0.446,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.24,
|
| 495 |
+
"learning_rate": 4.694720251945298e-06,
|
| 496 |
+
"logits/chosen": 0.49350661039352417,
|
| 497 |
+
"logits/rejected": 0.49793070554733276,
|
| 498 |
+
"logps/chosen": -325.5235900878906,
|
| 499 |
+
"logps/rejected": -307.2831726074219,
|
| 500 |
+
"loss": 0.0485,
|
| 501 |
+
"rewards/accuracies": 0.625,
|
| 502 |
+
"rewards/chosen": -0.027102241292595863,
|
| 503 |
+
"rewards/margins": 0.029582733288407326,
|
| 504 |
+
"rewards/rejected": -0.05668497085571289,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.25,
|
| 509 |
+
"learning_rate": 4.661051123281528e-06,
|
| 510 |
+
"logits/chosen": 0.43795689940452576,
|
| 511 |
+
"logits/rejected": 0.4705297350883484,
|
| 512 |
+
"logps/chosen": -304.1239318847656,
|
| 513 |
+
"logps/rejected": -286.34521484375,
|
| 514 |
+
"loss": 0.0474,
|
| 515 |
+
"rewards/accuracies": 0.59375,
|
| 516 |
+
"rewards/chosen": -0.029134873300790787,
|
| 517 |
+
"rewards/margins": 0.018936321139335632,
|
| 518 |
+
"rewards/rejected": -0.04807119444012642,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.26,
|
| 523 |
+
"learning_rate": 4.6257552230717536e-06,
|
| 524 |
+
"logits/chosen": 0.372799813747406,
|
| 525 |
+
"logits/rejected": 0.4166027903556824,
|
| 526 |
+
"logps/chosen": -288.1565856933594,
|
| 527 |
+
"logps/rejected": -327.458251953125,
|
| 528 |
+
"loss": 0.0417,
|
| 529 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 530 |
+
"rewards/chosen": -0.03428940102458,
|
| 531 |
+
"rewards/margins": 0.02542264387011528,
|
| 532 |
+
"rewards/rejected": -0.05971204489469528,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.27,
|
| 537 |
+
"learning_rate": 4.588859120963282e-06,
|
| 538 |
+
"logits/chosen": 0.27427417039871216,
|
| 539 |
+
"logits/rejected": 0.2968738377094269,
|
| 540 |
+
"logps/chosen": -267.9536437988281,
|
| 541 |
+
"logps/rejected": -281.6087646484375,
|
| 542 |
+
"loss": 0.049,
|
| 543 |
+
"rewards/accuracies": 0.59375,
|
| 544 |
+
"rewards/chosen": -0.03779522702097893,
|
| 545 |
+
"rewards/margins": 0.023776082322001457,
|
| 546 |
+
"rewards/rejected": -0.06157131865620613,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.27,
|
| 551 |
+
"learning_rate": 4.5503905911852435e-06,
|
| 552 |
+
"logits/chosen": 0.2773832678794861,
|
| 553 |
+
"logits/rejected": 0.28696539998054504,
|
| 554 |
+
"logps/chosen": -314.53521728515625,
|
| 555 |
+
"logps/rejected": -290.0928039550781,
|
| 556 |
+
"loss": 0.0493,
|
| 557 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 558 |
+
"rewards/chosen": -0.0507054328918457,
|
| 559 |
+
"rewards/margins": 0.023872187361121178,
|
| 560 |
+
"rewards/rejected": -0.07457762211561203,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.28,
|
| 565 |
+
"learning_rate": 4.510378591641036e-06,
|
| 566 |
+
"logits/chosen": 0.1799338012933731,
|
| 567 |
+
"logits/rejected": 0.21671123802661896,
|
| 568 |
+
"logps/chosen": -355.8804626464844,
|
| 569 |
+
"logps/rejected": -375.59918212890625,
|
| 570 |
+
"loss": 0.0388,
|
| 571 |
+
"rewards/accuracies": 0.606249988079071,
|
| 572 |
+
"rewards/chosen": -0.05081228166818619,
|
| 573 |
+
"rewards/margins": 0.030371317639946938,
|
| 574 |
+
"rewards/rejected": -0.08118359744548798,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.29,
|
| 579 |
+
"learning_rate": 4.468853242109712e-06,
|
| 580 |
+
"logits/chosen": 0.19799312949180603,
|
| 581 |
+
"logits/rejected": 0.1859453022480011,
|
| 582 |
+
"logps/chosen": -308.9555969238281,
|
| 583 |
+
"logps/rejected": -318.88787841796875,
|
| 584 |
+
"loss": 0.0442,
|
| 585 |
+
"rewards/accuracies": 0.637499988079071,
|
| 586 |
+
"rewards/chosen": -0.04206562787294388,
|
| 587 |
+
"rewards/margins": 0.02774595282971859,
|
| 588 |
+
"rewards/rejected": -0.06981157511472702,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.3,
|
| 593 |
+
"learning_rate": 4.42584580157276e-06,
|
| 594 |
+
"logits/chosen": 0.2413104772567749,
|
| 595 |
+
"logits/rejected": 0.3254660964012146,
|
| 596 |
+
"logps/chosen": -334.237548828125,
|
| 597 |
+
"logps/rejected": -312.25518798828125,
|
| 598 |
+
"loss": 0.0446,
|
| 599 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 600 |
+
"rewards/chosen": -0.037926387041807175,
|
| 601 |
+
"rewards/margins": 0.02587224170565605,
|
| 602 |
+
"rewards/rejected": -0.06379862874746323,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.31,
|
| 607 |
+
"learning_rate": 4.381388644683317e-06,
|
| 608 |
+
"logits/chosen": 0.299532949924469,
|
| 609 |
+
"logits/rejected": 0.3157830536365509,
|
| 610 |
+
"logps/chosen": -323.3116149902344,
|
| 611 |
+
"logps/rejected": -300.4678649902344,
|
| 612 |
+
"loss": 0.0474,
|
| 613 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 614 |
+
"rewards/chosen": -0.02991255559027195,
|
| 615 |
+
"rewards/margins": 0.02541544660925865,
|
| 616 |
+
"rewards/rejected": -0.05532800033688545,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.31,
|
| 621 |
+
"learning_rate": 4.33551523739555e-06,
|
| 622 |
+
"logits/chosen": 0.34920087456703186,
|
| 623 |
+
"logits/rejected": 0.4129624366760254,
|
| 624 |
+
"logps/chosen": -321.7519836425781,
|
| 625 |
+
"logps/rejected": -281.9065246582031,
|
| 626 |
+
"loss": 0.0467,
|
| 627 |
+
"rewards/accuracies": 0.668749988079071,
|
| 628 |
+
"rewards/chosen": -0.03106621466577053,
|
| 629 |
+
"rewards/margins": 0.026955511420965195,
|
| 630 |
+
"rewards/rejected": -0.058021724224090576,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.31,
|
| 635 |
+
"eval_logits/chosen": 0.3071232736110687,
|
| 636 |
+
"eval_logits/rejected": 0.34567520022392273,
|
| 637 |
+
"eval_logps/chosen": -306.9801940917969,
|
| 638 |
+
"eval_logps/rejected": -306.72412109375,
|
| 639 |
+
"eval_loss": 0.04428274184465408,
|
| 640 |
+
"eval_rewards/accuracies": 0.6032934188842773,
|
| 641 |
+
"eval_rewards/chosen": -0.03701169043779373,
|
| 642 |
+
"eval_rewards/margins": 0.021273676306009293,
|
| 643 |
+
"eval_rewards/rejected": -0.058285363018512726,
|
| 644 |
+
"eval_runtime": 374.0722,
|
| 645 |
+
"eval_samples_per_second": 5.347,
|
| 646 |
+
"eval_steps_per_second": 0.446,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.32,
|
| 651 |
+
"learning_rate": 4.288260111772535e-06,
|
| 652 |
+
"logits/chosen": 0.37154802680015564,
|
| 653 |
+
"logits/rejected": 0.3885091245174408,
|
| 654 |
+
"logps/chosen": -304.3981018066406,
|
| 655 |
+
"logps/rejected": -325.0479431152344,
|
| 656 |
+
"loss": 0.0471,
|
| 657 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 658 |
+
"rewards/chosen": -0.02885431982576847,
|
| 659 |
+
"rewards/margins": 0.03877053037285805,
|
| 660 |
+
"rewards/rejected": -0.06762484461069107,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.33,
|
| 665 |
+
"learning_rate": 4.239658839991594e-06,
|
| 666 |
+
"logits/chosen": 0.2403000146150589,
|
| 667 |
+
"logits/rejected": 0.3043418824672699,
|
| 668 |
+
"logps/chosen": -321.8272399902344,
|
| 669 |
+
"logps/rejected": -314.7815246582031,
|
| 670 |
+
"loss": 0.0413,
|
| 671 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 672 |
+
"rewards/chosen": -0.03149569034576416,
|
| 673 |
+
"rewards/margins": 0.03325023502111435,
|
| 674 |
+
"rewards/rejected": -0.06474592536687851,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.34,
|
| 679 |
+
"learning_rate": 4.189748007566686e-06,
|
| 680 |
+
"logits/chosen": 0.2293013632297516,
|
| 681 |
+
"logits/rejected": 0.25423663854599,
|
| 682 |
+
"logps/chosen": -295.85174560546875,
|
| 683 |
+
"logps/rejected": -301.47723388671875,
|
| 684 |
+
"loss": 0.0489,
|
| 685 |
+
"rewards/accuracies": 0.581250011920929,
|
| 686 |
+
"rewards/chosen": -0.03744658827781677,
|
| 687 |
+
"rewards/margins": 0.021524634212255478,
|
| 688 |
+
"rewards/rejected": -0.05897121876478195,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.35,
|
| 693 |
+
"learning_rate": 4.138565185807972e-06,
|
| 694 |
+
"logits/chosen": 0.2203875482082367,
|
| 695 |
+
"logits/rejected": 0.18498417735099792,
|
| 696 |
+
"logps/chosen": -288.0199890136719,
|
| 697 |
+
"logps/rejected": -322.2027282714844,
|
| 698 |
+
"loss": 0.0492,
|
| 699 |
+
"rewards/accuracies": 0.5625,
|
| 700 |
+
"rewards/chosen": -0.04328416287899017,
|
| 701 |
+
"rewards/margins": 0.039459649473428726,
|
| 702 |
+
"rewards/rejected": -0.0827438086271286,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.35,
|
| 707 |
+
"learning_rate": 4.086148903539311e-06,
|
| 708 |
+
"logits/chosen": 0.22477002441883087,
|
| 709 |
+
"logits/rejected": 0.24873077869415283,
|
| 710 |
+
"logps/chosen": -316.47052001953125,
|
| 711 |
+
"logps/rejected": -299.12127685546875,
|
| 712 |
+
"loss": 0.0436,
|
| 713 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 714 |
+
"rewards/chosen": -0.04419877380132675,
|
| 715 |
+
"rewards/margins": 0.02829798124730587,
|
| 716 |
+
"rewards/rejected": -0.07249675691127777,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.36,
|
| 721 |
+
"learning_rate": 4.032538618094972e-06,
|
| 722 |
+
"logits/chosen": 0.1782715618610382,
|
| 723 |
+
"logits/rejected": 0.22016020119190216,
|
| 724 |
+
"logps/chosen": -294.3050842285156,
|
| 725 |
+
"logps/rejected": -309.0475769042969,
|
| 726 |
+
"loss": 0.0393,
|
| 727 |
+
"rewards/accuracies": 0.668749988079071,
|
| 728 |
+
"rewards/chosen": -0.03711239993572235,
|
| 729 |
+
"rewards/margins": 0.03650590404868126,
|
| 730 |
+
"rewards/rejected": -0.07361830770969391,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.37,
|
| 735 |
+
"learning_rate": 3.977774685617386e-06,
|
| 736 |
+
"logits/chosen": 0.13345113396644592,
|
| 737 |
+
"logits/rejected": 0.18138858675956726,
|
| 738 |
+
"logps/chosen": -350.42303466796875,
|
| 739 |
+
"logps/rejected": -348.36273193359375,
|
| 740 |
+
"loss": 0.0353,
|
| 741 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 742 |
+
"rewards/chosen": -0.04365909844636917,
|
| 743 |
+
"rewards/margins": 0.03271768242120743,
|
| 744 |
+
"rewards/rejected": -0.0763767808675766,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.38,
|
| 749 |
+
"learning_rate": 3.92189833067831e-06,
|
| 750 |
+
"logits/chosen": 0.13094931840896606,
|
| 751 |
+
"logits/rejected": 0.17935091257095337,
|
| 752 |
+
"logps/chosen": -340.19708251953125,
|
| 753 |
+
"logps/rejected": -315.94659423828125,
|
| 754 |
+
"loss": 0.0415,
|
| 755 |
+
"rewards/accuracies": 0.675000011920929,
|
| 756 |
+
"rewards/chosen": -0.045614853501319885,
|
| 757 |
+
"rewards/margins": 0.03917178884148598,
|
| 758 |
+
"rewards/rejected": -0.08478663861751556,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.38,
|
| 763 |
+
"learning_rate": 3.864951615246261e-06,
|
| 764 |
+
"logits/chosen": 0.1737602949142456,
|
| 765 |
+
"logits/rejected": 0.2314929962158203,
|
| 766 |
+
"logps/chosen": -344.6983947753906,
|
| 767 |
+
"logps/rejected": -356.94989013671875,
|
| 768 |
+
"loss": 0.0406,
|
| 769 |
+
"rewards/accuracies": 0.625,
|
| 770 |
+
"rewards/chosen": -0.04962370544672012,
|
| 771 |
+
"rewards/margins": 0.04535280913114548,
|
| 772 |
+
"rewards/rejected": -0.0949765145778656,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.39,
|
| 777 |
+
"learning_rate": 3.806977407023581e-06,
|
| 778 |
+
"logits/chosen": 0.17347046732902527,
|
| 779 |
+
"logits/rejected": 0.15172746777534485,
|
| 780 |
+
"logps/chosen": -297.4417419433594,
|
| 781 |
+
"logps/rejected": -321.3406677246094,
|
| 782 |
+
"loss": 0.0359,
|
| 783 |
+
"rewards/accuracies": 0.637499988079071,
|
| 784 |
+
"rewards/chosen": -0.04872705414891243,
|
| 785 |
+
"rewards/margins": 0.05000175163149834,
|
| 786 |
+
"rewards/rejected": -0.09872879832983017,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.39,
|
| 791 |
+
"eval_logits/chosen": 0.1834835261106491,
|
| 792 |
+
"eval_logits/rejected": 0.21186047792434692,
|
| 793 |
+
"eval_logps/chosen": -327.32745361328125,
|
| 794 |
+
"eval_logps/rejected": -335.3609313964844,
|
| 795 |
+
"eval_loss": 0.042766325175762177,
|
| 796 |
+
"eval_rewards/accuracies": 0.598802387714386,
|
| 797 |
+
"eval_rewards/chosen": -0.05735894665122032,
|
| 798 |
+
"eval_rewards/margins": 0.029563238844275475,
|
| 799 |
+
"eval_rewards/rejected": -0.08692218363285065,
|
| 800 |
+
"eval_runtime": 374.1933,
|
| 801 |
+
"eval_samples_per_second": 5.345,
|
| 802 |
+
"eval_steps_per_second": 0.446,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.4,
|
| 807 |
+
"learning_rate": 3.7480193471769815e-06,
|
| 808 |
+
"logits/chosen": 0.1817350536584854,
|
| 809 |
+
"logits/rejected": 0.1976645439863205,
|
| 810 |
+
"logps/chosen": -316.01422119140625,
|
| 811 |
+
"logps/rejected": -315.81463623046875,
|
| 812 |
+
"loss": 0.0487,
|
| 813 |
+
"rewards/accuracies": 0.581250011920929,
|
| 814 |
+
"rewards/chosen": -0.04795747622847557,
|
| 815 |
+
"rewards/margins": 0.0256235059350729,
|
| 816 |
+
"rewards/rejected": -0.07358097285032272,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.41,
|
| 821 |
+
"learning_rate": 3.6881218174858354e-06,
|
| 822 |
+
"logits/chosen": 0.22255906462669373,
|
| 823 |
+
"logits/rejected": 0.2185302972793579,
|
| 824 |
+
"logps/chosen": -336.4140930175781,
|
| 825 |
+
"logps/rejected": -345.5931091308594,
|
| 826 |
+
"loss": 0.0464,
|
| 827 |
+
"rewards/accuracies": 0.668749988079071,
|
| 828 |
+
"rewards/chosen": -0.04861742630600929,
|
| 829 |
+
"rewards/margins": 0.034276556223630905,
|
| 830 |
+
"rewards/rejected": -0.0828939825296402,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.42,
|
| 835 |
+
"learning_rate": 3.627329906932964e-06,
|
| 836 |
+
"logits/chosen": 0.2149231731891632,
|
| 837 |
+
"logits/rejected": 0.23484310507774353,
|
| 838 |
+
"logps/chosen": -358.2276611328125,
|
| 839 |
+
"logps/rejected": -330.24505615234375,
|
| 840 |
+
"loss": 0.0425,
|
| 841 |
+
"rewards/accuracies": 0.65625,
|
| 842 |
+
"rewards/chosen": -0.05005684494972229,
|
| 843 |
+
"rewards/margins": 0.03168889507651329,
|
| 844 |
+
"rewards/rejected": -0.08174573630094528,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 0.42,
|
| 849 |
+
"learning_rate": 3.5656893777630686e-06,
|
| 850 |
+
"logits/chosen": 0.2030162811279297,
|
| 851 |
+
"logits/rejected": 0.25424039363861084,
|
| 852 |
+
"logps/chosen": -298.22308349609375,
|
| 853 |
+
"logps/rejected": -316.8375549316406,
|
| 854 |
+
"loss": 0.0439,
|
| 855 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 856 |
+
"rewards/chosen": -0.051857758313417435,
|
| 857 |
+
"rewards/margins": 0.04128115624189377,
|
| 858 |
+
"rewards/rejected": -0.0931389182806015,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 0.43,
|
| 863 |
+
"learning_rate": 3.503246631034345e-06,
|
| 864 |
+
"logits/chosen": 0.2722618579864502,
|
| 865 |
+
"logits/rejected": 0.3168015778064728,
|
| 866 |
+
"logps/chosen": -299.116943359375,
|
| 867 |
+
"logps/rejected": -307.1891784667969,
|
| 868 |
+
"loss": 0.0448,
|
| 869 |
+
"rewards/accuracies": 0.625,
|
| 870 |
+
"rewards/chosen": -0.04390637204051018,
|
| 871 |
+
"rewards/margins": 0.0321880504488945,
|
| 872 |
+
"rewards/rejected": -0.07609442621469498,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.44,
|
| 877 |
+
"learning_rate": 3.440048671689219e-06,
|
| 878 |
+
"logits/chosen": 0.24053359031677246,
|
| 879 |
+
"logits/rejected": 0.27442485094070435,
|
| 880 |
+
"logps/chosen": -323.40631103515625,
|
| 881 |
+
"logps/rejected": -323.588623046875,
|
| 882 |
+
"loss": 0.0408,
|
| 883 |
+
"rewards/accuracies": 0.637499988079071,
|
| 884 |
+
"rewards/chosen": -0.0459776408970356,
|
| 885 |
+
"rewards/margins": 0.036371566355228424,
|
| 886 |
+
"rewards/rejected": -0.08234919607639313,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 0.45,
|
| 891 |
+
"learning_rate": 3.3761430731705056e-06,
|
| 892 |
+
"logits/chosen": 0.25353458523750305,
|
| 893 |
+
"logits/rejected": 0.33455875515937805,
|
| 894 |
+
"logps/chosen": -323.89892578125,
|
| 895 |
+
"logps/rejected": -319.5647888183594,
|
| 896 |
+
"loss": 0.0347,
|
| 897 |
+
"rewards/accuracies": 0.625,
|
| 898 |
+
"rewards/chosen": -0.04967617988586426,
|
| 899 |
+
"rewards/margins": 0.028463760390877724,
|
| 900 |
+
"rewards/rejected": -0.07813994586467743,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 0.46,
|
| 905 |
+
"learning_rate": 3.311577941609604e-06,
|
| 906 |
+
"logits/chosen": 0.31304338574409485,
|
| 907 |
+
"logits/rejected": 0.3095194697380066,
|
| 908 |
+
"logps/chosen": -268.1946105957031,
|
| 909 |
+
"logps/rejected": -263.7663879394531,
|
| 910 |
+
"loss": 0.0451,
|
| 911 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 912 |
+
"rewards/chosen": -0.043435920029878616,
|
| 913 |
+
"rewards/margins": 0.024017784744501114,
|
| 914 |
+
"rewards/rejected": -0.06745370477437973,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.46,
|
| 919 |
+
"learning_rate": 3.2464018796137157e-06,
|
| 920 |
+
"logits/chosen": 0.2560151219367981,
|
| 921 |
+
"logits/rejected": 0.24582383036613464,
|
| 922 |
+
"logps/chosen": -306.5500183105469,
|
| 923 |
+
"logps/rejected": -292.2866516113281,
|
| 924 |
+
"loss": 0.0383,
|
| 925 |
+
"rewards/accuracies": 0.5625,
|
| 926 |
+
"rewards/chosen": -0.040644846856594086,
|
| 927 |
+
"rewards/margins": 0.024079840630292892,
|
| 928 |
+
"rewards/rejected": -0.06472468376159668,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.47,
|
| 933 |
+
"learning_rate": 3.1806639496793245e-06,
|
| 934 |
+
"logits/chosen": 0.29677996039390564,
|
| 935 |
+
"logits/rejected": 0.3049861490726471,
|
| 936 |
+
"logps/chosen": -307.2038269042969,
|
| 937 |
+
"logps/rejected": -326.469970703125,
|
| 938 |
+
"loss": 0.0431,
|
| 939 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 940 |
+
"rewards/chosen": -0.033155038952827454,
|
| 941 |
+
"rewards/margins": 0.02939048409461975,
|
| 942 |
+
"rewards/rejected": -0.0625455230474472,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 0.47,
|
| 947 |
+
"eval_logits/chosen": 0.2554236948490143,
|
| 948 |
+
"eval_logits/rejected": 0.28905177116394043,
|
| 949 |
+
"eval_logps/chosen": -314.9629821777344,
|
| 950 |
+
"eval_logps/rejected": -320.91607666015625,
|
| 951 |
+
"eval_loss": 0.041808027774095535,
|
| 952 |
+
"eval_rewards/accuracies": 0.6032934188842773,
|
| 953 |
+
"eval_rewards/chosen": -0.04499450698494911,
|
| 954 |
+
"eval_rewards/margins": 0.02748279832303524,
|
| 955 |
+
"eval_rewards/rejected": -0.0724773034453392,
|
| 956 |
+
"eval_runtime": 373.9667,
|
| 957 |
+
"eval_samples_per_second": 5.348,
|
| 958 |
+
"eval_steps_per_second": 0.447,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.48,
|
| 963 |
+
"learning_rate": 3.114413637259484e-06,
|
| 964 |
+
"logits/chosen": 0.2767940163612366,
|
| 965 |
+
"logits/rejected": 0.30368971824645996,
|
| 966 |
+
"logps/chosen": -312.98223876953125,
|
| 967 |
+
"logps/rejected": -334.64166259765625,
|
| 968 |
+
"loss": 0.0419,
|
| 969 |
+
"rewards/accuracies": 0.668749988079071,
|
| 970 |
+
"rewards/chosen": -0.040467552840709686,
|
| 971 |
+
"rewards/margins": 0.03511633723974228,
|
| 972 |
+
"rewards/rejected": -0.07558389008045197,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 0.49,
|
| 977 |
+
"learning_rate": 3.0477008135127247e-06,
|
| 978 |
+
"logits/chosen": 0.22974368929862976,
|
| 979 |
+
"logits/rejected": 0.2028408795595169,
|
| 980 |
+
"logps/chosen": -320.90545654296875,
|
| 981 |
+
"logps/rejected": -328.469970703125,
|
| 982 |
+
"loss": 0.0438,
|
| 983 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 984 |
+
"rewards/chosen": -0.047637082636356354,
|
| 985 |
+
"rewards/margins": 0.03967064619064331,
|
| 986 |
+
"rewards/rejected": -0.08730772882699966,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.49,
|
| 991 |
+
"learning_rate": 2.980575697761603e-06,
|
| 992 |
+
"logits/chosen": 0.25714489817619324,
|
| 993 |
+
"logits/rejected": 0.24402575194835663,
|
| 994 |
+
"logps/chosen": -298.93994140625,
|
| 995 |
+
"logps/rejected": -321.35748291015625,
|
| 996 |
+
"loss": 0.0466,
|
| 997 |
+
"rewards/accuracies": 0.581250011920929,
|
| 998 |
+
"rewards/chosen": -0.05835377424955368,
|
| 999 |
+
"rewards/margins": 0.031576722860336304,
|
| 1000 |
+
"rewards/rejected": -0.08993048965930939,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.5,
|
| 1005 |
+
"learning_rate": 2.9130888196891755e-06,
|
| 1006 |
+
"logits/chosen": 0.21983161568641663,
|
| 1007 |
+
"logits/rejected": 0.2652173936367035,
|
| 1008 |
+
"logps/chosen": -296.48626708984375,
|
| 1009 |
+
"logps/rejected": -289.70587158203125,
|
| 1010 |
+
"loss": 0.0448,
|
| 1011 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 1012 |
+
"rewards/chosen": -0.049575865268707275,
|
| 1013 |
+
"rewards/margins": 0.019396495074033737,
|
| 1014 |
+
"rewards/rejected": -0.06897236406803131,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.51,
|
| 1019 |
+
"learning_rate": 2.845290981301834e-06,
|
| 1020 |
+
"logits/chosen": 0.30579501390457153,
|
| 1021 |
+
"logits/rejected": 0.27916720509529114,
|
| 1022 |
+
"logps/chosen": -314.1757507324219,
|
| 1023 |
+
"logps/rejected": -312.4922180175781,
|
| 1024 |
+
"loss": 0.0377,
|
| 1025 |
+
"rewards/accuracies": 0.550000011920929,
|
| 1026 |
+
"rewards/chosen": -0.05070630833506584,
|
| 1027 |
+
"rewards/margins": 0.03026432916522026,
|
| 1028 |
+
"rewards/rejected": -0.0809706375002861,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.52,
|
| 1033 |
+
"learning_rate": 2.7772332186871464e-06,
|
| 1034 |
+
"logits/chosen": 0.2696114182472229,
|
| 1035 |
+
"logits/rejected": 0.30878639221191406,
|
| 1036 |
+
"logps/chosen": -332.66961669921875,
|
| 1037 |
+
"logps/rejected": -331.0030822753906,
|
| 1038 |
+
"loss": 0.0431,
|
| 1039 |
+
"rewards/accuracies": 0.625,
|
| 1040 |
+
"rewards/chosen": -0.05702507495880127,
|
| 1041 |
+
"rewards/margins": 0.0359366312623024,
|
| 1042 |
+
"rewards/rejected": -0.09296169131994247,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 0.53,
|
| 1047 |
+
"learning_rate": 2.708966763595493e-06,
|
| 1048 |
+
"logits/chosen": 0.2562243938446045,
|
| 1049 |
+
"logits/rejected": 0.2766057848930359,
|
| 1050 |
+
"logps/chosen": -330.1766357421875,
|
| 1051 |
+
"logps/rejected": -306.47894287109375,
|
| 1052 |
+
"loss": 0.0439,
|
| 1053 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1054 |
+
"rewards/chosen": -0.052916429936885834,
|
| 1055 |
+
"rewards/margins": 0.035131536424160004,
|
| 1056 |
+
"rewards/rejected": -0.08804796636104584,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.53,
|
| 1061 |
+
"learning_rate": 2.640543004874409e-06,
|
| 1062 |
+
"logits/chosen": 0.21568675339221954,
|
| 1063 |
+
"logits/rejected": 0.2647091746330261,
|
| 1064 |
+
"logps/chosen": -336.7883605957031,
|
| 1065 |
+
"logps/rejected": -329.66461181640625,
|
| 1066 |
+
"loss": 0.0448,
|
| 1067 |
+
"rewards/accuracies": 0.59375,
|
| 1068 |
+
"rewards/chosen": -0.050951045006513596,
|
| 1069 |
+
"rewards/margins": 0.03465459495782852,
|
| 1070 |
+
"rewards/rejected": -0.08560563623905182,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.54,
|
| 1075 |
+
"learning_rate": 2.572013449784671e-06,
|
| 1076 |
+
"logits/chosen": 0.2518898844718933,
|
| 1077 |
+
"logits/rejected": 0.22508414089679718,
|
| 1078 |
+
"logps/chosen": -295.5574645996094,
|
| 1079 |
+
"logps/rejected": -297.5489807128906,
|
| 1080 |
+
"loss": 0.0412,
|
| 1081 |
+
"rewards/accuracies": 0.5625,
|
| 1082 |
+
"rewards/chosen": -0.0494476854801178,
|
| 1083 |
+
"rewards/margins": 0.02454470470547676,
|
| 1084 |
+
"rewards/rejected": -0.07399239391088486,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.55,
|
| 1089 |
+
"learning_rate": 2.503429685227245e-06,
|
| 1090 |
+
"logits/chosen": 0.20451202988624573,
|
| 1091 |
+
"logits/rejected": 0.21218447387218475,
|
| 1092 |
+
"logps/chosen": -361.32379150390625,
|
| 1093 |
+
"logps/rejected": -326.19989013671875,
|
| 1094 |
+
"loss": 0.0438,
|
| 1095 |
+
"rewards/accuracies": 0.625,
|
| 1096 |
+
"rewards/chosen": -0.05410303920507431,
|
| 1097 |
+
"rewards/margins": 0.023143690079450607,
|
| 1098 |
+
"rewards/rejected": -0.07724673300981522,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 0.55,
|
| 1103 |
+
"eval_logits/chosen": 0.204001784324646,
|
| 1104 |
+
"eval_logits/rejected": 0.23564448952674866,
|
| 1105 |
+
"eval_logps/chosen": -327.3254089355469,
|
| 1106 |
+
"eval_logps/rejected": -337.3518981933594,
|
| 1107 |
+
"eval_loss": 0.04129507392644882,
|
| 1108 |
+
"eval_rewards/accuracies": 0.6017963886260986,
|
| 1109 |
+
"eval_rewards/chosen": -0.05735692009329796,
|
| 1110 |
+
"eval_rewards/margins": 0.03155623376369476,
|
| 1111 |
+
"eval_rewards/rejected": -0.08891315013170242,
|
| 1112 |
+
"eval_runtime": 374.0567,
|
| 1113 |
+
"eval_samples_per_second": 5.347,
|
| 1114 |
+
"eval_steps_per_second": 0.446,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.56,
|
| 1119 |
+
"learning_rate": 2.434843338910286e-06,
|
| 1120 |
+
"logits/chosen": 0.1988871991634369,
|
| 1121 |
+
"logits/rejected": 0.2530394494533539,
|
| 1122 |
+
"logps/chosen": -362.696533203125,
|
| 1123 |
+
"logps/rejected": -351.73992919921875,
|
| 1124 |
+
"loss": 0.039,
|
| 1125 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1126 |
+
"rewards/chosen": -0.04888119548559189,
|
| 1127 |
+
"rewards/margins": 0.0342581570148468,
|
| 1128 |
+
"rewards/rejected": -0.0831393450498581,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.57,
|
| 1133 |
+
"learning_rate": 2.3663060404854155e-06,
|
| 1134 |
+
"logits/chosen": 0.19785156846046448,
|
| 1135 |
+
"logits/rejected": 0.2040957510471344,
|
| 1136 |
+
"logps/chosen": -303.4078674316406,
|
| 1137 |
+
"logps/rejected": -326.0086669921875,
|
| 1138 |
+
"loss": 0.0398,
|
| 1139 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1140 |
+
"rewards/chosen": -0.05251380056142807,
|
| 1141 |
+
"rewards/margins": 0.047541745007038116,
|
| 1142 |
+
"rewards/rejected": -0.10005555301904678,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.57,
|
| 1147 |
+
"learning_rate": 2.2978693826825406e-06,
|
| 1148 |
+
"logits/chosen": 0.21812646090984344,
|
| 1149 |
+
"logits/rejected": 0.2294880896806717,
|
| 1150 |
+
"logps/chosen": -347.4920959472656,
|
| 1151 |
+
"logps/rejected": -328.907958984375,
|
| 1152 |
+
"loss": 0.0374,
|
| 1153 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1154 |
+
"rewards/chosen": -0.05341614410281181,
|
| 1155 |
+
"rewards/margins": 0.031886156648397446,
|
| 1156 |
+
"rewards/rejected": -0.08530230075120926,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.58,
|
| 1161 |
+
"learning_rate": 2.2295848824724612e-06,
|
| 1162 |
+
"logits/chosen": 0.21070751547813416,
|
| 1163 |
+
"logits/rejected": 0.25296124815940857,
|
| 1164 |
+
"logps/chosen": -295.4853210449219,
|
| 1165 |
+
"logps/rejected": -316.27044677734375,
|
| 1166 |
+
"loss": 0.0525,
|
| 1167 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1168 |
+
"rewards/chosen": -0.05813244730234146,
|
| 1169 |
+
"rewards/margins": 0.0497218519449234,
|
| 1170 |
+
"rewards/rejected": -0.10785429179668427,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.59,
|
| 1175 |
+
"learning_rate": 2.1615039422865136e-06,
|
| 1176 |
+
"logits/chosen": 0.25933319330215454,
|
| 1177 |
+
"logits/rejected": 0.31064844131469727,
|
| 1178 |
+
"logps/chosen": -309.44964599609375,
|
| 1179 |
+
"logps/rejected": -307.4235534667969,
|
| 1180 |
+
"loss": 0.0474,
|
| 1181 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1182 |
+
"rewards/chosen": -0.05073714256286621,
|
| 1183 |
+
"rewards/margins": 0.03562631830573082,
|
| 1184 |
+
"rewards/rejected": -0.08636346459388733,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.6,
|
| 1189 |
+
"learning_rate": 2.0936778113224253e-06,
|
| 1190 |
+
"logits/chosen": 0.3158121109008789,
|
| 1191 |
+
"logits/rejected": 0.31753307580947876,
|
| 1192 |
+
"logps/chosen": -302.07086181640625,
|
| 1193 |
+
"logps/rejected": -300.81793212890625,
|
| 1194 |
+
"loss": 0.0404,
|
| 1195 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1196 |
+
"rewards/chosen": -0.04117770120501518,
|
| 1197 |
+
"rewards/margins": 0.04220277816057205,
|
| 1198 |
+
"rewards/rejected": -0.08338047564029694,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.6,
|
| 1203 |
+
"learning_rate": 2.0261575469655304e-06,
|
| 1204 |
+
"logits/chosen": 0.2294500768184662,
|
| 1205 |
+
"logits/rejected": 0.2536849081516266,
|
| 1206 |
+
"logps/chosen": -320.7763671875,
|
| 1207 |
+
"logps/rejected": -321.6782531738281,
|
| 1208 |
+
"loss": 0.0389,
|
| 1209 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1210 |
+
"rewards/chosen": -0.053652387112379074,
|
| 1211 |
+
"rewards/margins": 0.030662674456834793,
|
| 1212 |
+
"rewards/rejected": -0.08431507647037506,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.61,
|
| 1217 |
+
"learning_rate": 1.9589939763543693e-06,
|
| 1218 |
+
"logits/chosen": 0.26485732197761536,
|
| 1219 |
+
"logits/rejected": 0.21588650345802307,
|
| 1220 |
+
"logps/chosen": -304.3012390136719,
|
| 1221 |
+
"logps/rejected": -291.78192138671875,
|
| 1222 |
+
"loss": 0.0431,
|
| 1223 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1224 |
+
"rewards/chosen": -0.04734267294406891,
|
| 1225 |
+
"rewards/margins": 0.03097641095519066,
|
| 1226 |
+
"rewards/rejected": -0.07831908762454987,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.62,
|
| 1231 |
+
"learning_rate": 1.8922376581196107e-06,
|
| 1232 |
+
"logits/chosen": 0.28727906942367554,
|
| 1233 |
+
"logits/rejected": 0.27608275413513184,
|
| 1234 |
+
"logps/chosen": -343.40472412109375,
|
| 1235 |
+
"logps/rejected": -321.67230224609375,
|
| 1236 |
+
"loss": 0.0498,
|
| 1237 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1238 |
+
"rewards/chosen": -0.048335738480091095,
|
| 1239 |
+
"rewards/margins": 0.030372479930520058,
|
| 1240 |
+
"rewards/rejected": -0.078708216547966,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.63,
|
| 1245 |
+
"learning_rate": 1.8259388443250993e-06,
|
| 1246 |
+
"logits/chosen": 0.23254342377185822,
|
| 1247 |
+
"logits/rejected": 0.2663263976573944,
|
| 1248 |
+
"logps/chosen": -303.95733642578125,
|
| 1249 |
+
"logps/rejected": -291.7852478027344,
|
| 1250 |
+
"loss": 0.0446,
|
| 1251 |
+
"rewards/accuracies": 0.5625,
|
| 1252 |
+
"rewards/chosen": -0.03941676765680313,
|
| 1253 |
+
"rewards/margins": 0.02897232212126255,
|
| 1254 |
+
"rewards/rejected": -0.06838909536600113,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.63,
|
| 1259 |
+
"eval_logits/chosen": 0.22356104850769043,
|
| 1260 |
+
"eval_logits/rejected": 0.2566269636154175,
|
| 1261 |
+
"eval_logps/chosen": -322.177734375,
|
| 1262 |
+
"eval_logps/rejected": -332.6603088378906,
|
| 1263 |
+
"eval_loss": 0.040854789316654205,
|
| 1264 |
+
"eval_rewards/accuracies": 0.6047903895378113,
|
| 1265 |
+
"eval_rewards/chosen": -0.05220925062894821,
|
| 1266 |
+
"eval_rewards/margins": 0.03201233595609665,
|
| 1267 |
+
"eval_rewards/rejected": -0.08422157913446426,
|
| 1268 |
+
"eval_runtime": 374.1827,
|
| 1269 |
+
"eval_samples_per_second": 5.345,
|
| 1270 |
+
"eval_steps_per_second": 0.446,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 0.64,
|
| 1275 |
+
"learning_rate": 1.760147442639679e-06,
|
| 1276 |
+
"logits/chosen": 0.18028207123279572,
|
| 1277 |
+
"logits/rejected": 0.24531738460063934,
|
| 1278 |
+
"logps/chosen": -317.4986267089844,
|
| 1279 |
+
"logps/rejected": -286.6923522949219,
|
| 1280 |
+
"loss": 0.045,
|
| 1281 |
+
"rewards/accuracies": 0.625,
|
| 1282 |
+
"rewards/chosen": -0.04818979650735855,
|
| 1283 |
+
"rewards/margins": 0.03410783410072327,
|
| 1284 |
+
"rewards/rejected": -0.08229763805866241,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.64,
|
| 1289 |
+
"learning_rate": 1.6949129787682628e-06,
|
| 1290 |
+
"logits/chosen": 0.1836201250553131,
|
| 1291 |
+
"logits/rejected": 0.19242016971111298,
|
| 1292 |
+
"logps/chosen": -358.3663024902344,
|
| 1293 |
+
"logps/rejected": -320.7300720214844,
|
| 1294 |
+
"loss": 0.056,
|
| 1295 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1296 |
+
"rewards/chosen": -0.05281525105237961,
|
| 1297 |
+
"rewards/margins": 0.030187761411070824,
|
| 1298 |
+
"rewards/rejected": -0.08300301432609558,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.65,
|
| 1303 |
+
"learning_rate": 1.6302845591704348e-06,
|
| 1304 |
+
"logits/chosen": 0.23501524329185486,
|
| 1305 |
+
"logits/rejected": 0.253219872713089,
|
| 1306 |
+
"logps/chosen": -301.3849182128906,
|
| 1307 |
+
"logps/rejected": -324.7459716796875,
|
| 1308 |
+
"loss": 0.0483,
|
| 1309 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1310 |
+
"rewards/chosen": -0.047373272478580475,
|
| 1311 |
+
"rewards/margins": 0.03554272651672363,
|
| 1312 |
+
"rewards/rejected": -0.08291599899530411,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.66,
|
| 1317 |
+
"learning_rate": 1.5663108340946465e-06,
|
| 1318 |
+
"logits/chosen": 0.27467894554138184,
|
| 1319 |
+
"logits/rejected": 0.29102447628974915,
|
| 1320 |
+
"logps/chosen": -297.8023376464844,
|
| 1321 |
+
"logps/rejected": -317.1583251953125,
|
| 1322 |
+
"loss": 0.0398,
|
| 1323 |
+
"rewards/accuracies": 0.59375,
|
| 1324 |
+
"rewards/chosen": -0.0461508110165596,
|
| 1325 |
+
"rewards/margins": 0.037653181701898575,
|
| 1326 |
+
"rewards/rejected": -0.08380399644374847,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.67,
|
| 1331 |
+
"learning_rate": 1.5030399609558364e-06,
|
| 1332 |
+
"logits/chosen": 0.22526296973228455,
|
| 1333 |
+
"logits/rejected": 0.2629440426826477,
|
| 1334 |
+
"logps/chosen": -314.87005615234375,
|
| 1335 |
+
"logps/rejected": -327.7728576660156,
|
| 1336 |
+
"loss": 0.0462,
|
| 1337 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1338 |
+
"rewards/chosen": -0.05448513478040695,
|
| 1339 |
+
"rewards/margins": 0.03063533641397953,
|
| 1340 |
+
"rewards/rejected": -0.08512047678232193,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.68,
|
| 1345 |
+
"learning_rate": 1.4405195680840357e-06,
|
| 1346 |
+
"logits/chosen": 0.21215219795703888,
|
| 1347 |
+
"logits/rejected": 0.24855566024780273,
|
| 1348 |
+
"logps/chosen": -299.90167236328125,
|
| 1349 |
+
"logps/rejected": -298.214111328125,
|
| 1350 |
+
"loss": 0.0409,
|
| 1351 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 1352 |
+
"rewards/chosen": -0.04720069095492363,
|
| 1353 |
+
"rewards/margins": 0.04011126235127449,
|
| 1354 |
+
"rewards/rejected": -0.08731194585561752,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 0.68,
|
| 1359 |
+
"learning_rate": 1.378796718871252e-06,
|
| 1360 |
+
"logits/chosen": 0.22330372035503387,
|
| 1361 |
+
"logits/rejected": 0.27839145064353943,
|
| 1362 |
+
"logps/chosen": -352.99810791015625,
|
| 1363 |
+
"logps/rejected": -318.991943359375,
|
| 1364 |
+
"loss": 0.0429,
|
| 1365 |
+
"rewards/accuracies": 0.625,
|
| 1366 |
+
"rewards/chosen": -0.04801223799586296,
|
| 1367 |
+
"rewards/margins": 0.03813933581113815,
|
| 1368 |
+
"rewards/rejected": -0.08615156263113022,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 0.69,
|
| 1373 |
+
"learning_rate": 1.3179178763436302e-06,
|
| 1374 |
+
"logits/chosen": 0.2164164036512375,
|
| 1375 |
+
"logits/rejected": 0.2723962664604187,
|
| 1376 |
+
"logps/chosen": -296.51507568359375,
|
| 1377 |
+
"logps/rejected": -311.3013610839844,
|
| 1378 |
+
"loss": 0.0331,
|
| 1379 |
+
"rewards/accuracies": 0.5625,
|
| 1380 |
+
"rewards/chosen": -0.045425109565258026,
|
| 1381 |
+
"rewards/margins": 0.043222926557064056,
|
| 1382 |
+
"rewards/rejected": -0.08864804357290268,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.7,
|
| 1387 |
+
"learning_rate": 1.2579288681855364e-06,
|
| 1388 |
+
"logits/chosen": 0.2106570303440094,
|
| 1389 |
+
"logits/rejected": 0.20854637026786804,
|
| 1390 |
+
"logps/chosen": -300.02044677734375,
|
| 1391 |
+
"logps/rejected": -311.44976806640625,
|
| 1392 |
+
"loss": 0.0433,
|
| 1393 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1394 |
+
"rewards/chosen": -0.051221005618572235,
|
| 1395 |
+
"rewards/margins": 0.03814007714390755,
|
| 1396 |
+
"rewards/rejected": -0.08936108648777008,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.71,
|
| 1401 |
+
"learning_rate": 1.1988748522419163e-06,
|
| 1402 |
+
"logits/chosen": 0.23926115036010742,
|
| 1403 |
+
"logits/rejected": 0.26336807012557983,
|
| 1404 |
+
"logps/chosen": -332.501953125,
|
| 1405 |
+
"logps/rejected": -341.2127685546875,
|
| 1406 |
+
"loss": 0.0426,
|
| 1407 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1408 |
+
"rewards/chosen": -0.055366236716508865,
|
| 1409 |
+
"rewards/margins": 0.04208333045244217,
|
| 1410 |
+
"rewards/rejected": -0.09744956344366074,
|
| 1411 |
+
"step": 900
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 0.71,
|
| 1415 |
+
"eval_logits/chosen": 0.17968809604644775,
|
| 1416 |
+
"eval_logits/rejected": 0.20887503027915955,
|
| 1417 |
+
"eval_logps/chosen": -332.3423767089844,
|
| 1418 |
+
"eval_logps/rejected": -346.1493835449219,
|
| 1419 |
+
"eval_loss": 0.04077794775366783,
|
| 1420 |
+
"eval_rewards/accuracies": 0.6047903895378113,
|
| 1421 |
+
"eval_rewards/chosen": -0.06237388774752617,
|
| 1422 |
+
"eval_rewards/margins": 0.03533677011728287,
|
| 1423 |
+
"eval_rewards/rejected": -0.09771064668893814,
|
| 1424 |
+
"eval_runtime": 374.5567,
|
| 1425 |
+
"eval_samples_per_second": 5.34,
|
| 1426 |
+
"eval_steps_per_second": 0.446,
|
| 1427 |
+
"step": 900
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.71,
|
| 1431 |
+
"learning_rate": 1.1408002825248842e-06,
|
| 1432 |
+
"logits/chosen": 0.17859011888504028,
|
| 1433 |
+
"logits/rejected": 0.1722099781036377,
|
| 1434 |
+
"logps/chosen": -349.0040588378906,
|
| 1435 |
+
"logps/rejected": -347.06195068359375,
|
| 1436 |
+
"loss": 0.036,
|
| 1437 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1438 |
+
"rewards/chosen": -0.055424802005290985,
|
| 1439 |
+
"rewards/margins": 0.03749425336718559,
|
| 1440 |
+
"rewards/rejected": -0.09291905909776688,
|
| 1441 |
+
"step": 910
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 0.72,
|
| 1445 |
+
"learning_rate": 1.0837488757501369e-06,
|
| 1446 |
+
"logits/chosen": 0.1789877861738205,
|
| 1447 |
+
"logits/rejected": 0.2344355285167694,
|
| 1448 |
+
"logps/chosen": -317.1181640625,
|
| 1449 |
+
"logps/rejected": -326.86566162109375,
|
| 1450 |
+
"loss": 0.039,
|
| 1451 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1452 |
+
"rewards/chosen": -0.053631655871868134,
|
| 1453 |
+
"rewards/margins": 0.03638458251953125,
|
| 1454 |
+
"rewards/rejected": -0.09001623094081879,
|
| 1455 |
+
"step": 920
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 0.73,
|
| 1459 |
+
"learning_rate": 1.027763578428379e-06,
|
| 1460 |
+
"logits/chosen": 0.18564803898334503,
|
| 1461 |
+
"logits/rejected": 0.1936606466770172,
|
| 1462 |
+
"logps/chosen": -293.61920166015625,
|
| 1463 |
+
"logps/rejected": -302.94744873046875,
|
| 1464 |
+
"loss": 0.0454,
|
| 1465 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1466 |
+
"rewards/chosen": -0.06316958367824554,
|
| 1467 |
+
"rewards/margins": 0.03379397466778755,
|
| 1468 |
+
"rewards/rejected": -0.09696356952190399,
|
| 1469 |
+
"step": 930
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 0.74,
|
| 1473 |
+
"learning_rate": 9.728865345365379e-07,
|
| 1474 |
+
"logits/chosen": 0.19354644417762756,
|
| 1475 |
+
"logits/rejected": 0.21065323054790497,
|
| 1476 |
+
"logps/chosen": -340.0384826660156,
|
| 1477 |
+
"logps/rejected": -319.240966796875,
|
| 1478 |
+
"loss": 0.0495,
|
| 1479 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1480 |
+
"rewards/chosen": -0.05120069906115532,
|
| 1481 |
+
"rewards/margins": 0.04345576837658882,
|
| 1482 |
+
"rewards/rejected": -0.09465646743774414,
|
| 1483 |
+
"step": 940
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 0.75,
|
| 1487 |
+
"learning_rate": 9.191590537930975e-07,
|
| 1488 |
+
"logits/chosen": 0.169469952583313,
|
| 1489 |
+
"logits/rejected": 0.20710797607898712,
|
| 1490 |
+
"logps/chosen": -331.83740234375,
|
| 1491 |
+
"logps/rejected": -317.3833312988281,
|
| 1492 |
+
"loss": 0.0433,
|
| 1493 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1494 |
+
"rewards/chosen": -0.04739029332995415,
|
| 1495 |
+
"rewards/margins": 0.049665920436382294,
|
| 1496 |
+
"rewards/rejected": -0.09705621004104614,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 0.75,
|
| 1501 |
+
"learning_rate": 8.666215805614373e-07,
|
| 1502 |
+
"logits/chosen": 0.17803916335105896,
|
| 1503 |
+
"logits/rejected": 0.24121513962745667,
|
| 1504 |
+
"logps/chosen": -340.6368713378906,
|
| 1505 |
+
"logps/rejected": -323.7513732910156,
|
| 1506 |
+
"loss": 0.0448,
|
| 1507 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1508 |
+
"rewards/chosen": -0.06622298806905746,
|
| 1509 |
+
"rewards/margins": 0.023390918970108032,
|
| 1510 |
+
"rewards/rejected": -0.0896139144897461,
|
| 1511 |
+
"step": 960
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 0.76,
|
| 1515 |
+
"learning_rate": 8.153136634045844e-07,
|
| 1516 |
+
"logits/chosen": 0.19488176703453064,
|
| 1517 |
+
"logits/rejected": 0.2004108428955078,
|
| 1518 |
+
"logps/chosen": -308.45074462890625,
|
| 1519 |
+
"logps/rejected": -309.8042297363281,
|
| 1520 |
+
"loss": 0.0388,
|
| 1521 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1522 |
+
"rewards/chosen": -0.05112530663609505,
|
| 1523 |
+
"rewards/margins": 0.04051050543785095,
|
| 1524 |
+
"rewards/rejected": -0.0916358157992363,
|
| 1525 |
+
"step": 970
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.77,
|
| 1529 |
+
"learning_rate": 7.652739253142915e-07,
|
| 1530 |
+
"logits/chosen": 0.24685409665107727,
|
| 1531 |
+
"logits/rejected": 0.24745607376098633,
|
| 1532 |
+
"logps/chosen": -322.38018798828125,
|
| 1533 |
+
"logps/rejected": -314.3333435058594,
|
| 1534 |
+
"loss": 0.0421,
|
| 1535 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1536 |
+
"rewards/chosen": -0.05213675647974014,
|
| 1537 |
+
"rewards/margins": 0.03133903816342354,
|
| 1538 |
+
"rewards/rejected": -0.08347579836845398,
|
| 1539 |
+
"step": 980
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.78,
|
| 1543 |
+
"learning_rate": 7.165400346368648e-07,
|
| 1544 |
+
"logits/chosen": 0.2231186181306839,
|
| 1545 |
+
"logits/rejected": 0.2616454064846039,
|
| 1546 |
+
"logps/chosen": -344.0087890625,
|
| 1547 |
+
"logps/rejected": -308.85052490234375,
|
| 1548 |
+
"loss": 0.0467,
|
| 1549 |
+
"rewards/accuracies": 0.65625,
|
| 1550 |
+
"rewards/chosen": -0.05246872827410698,
|
| 1551 |
+
"rewards/margins": 0.03322756290435791,
|
| 1552 |
+
"rewards/rejected": -0.08569629490375519,
|
| 1553 |
+
"step": 990
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 0.79,
|
| 1557 |
+
"learning_rate": 6.691486767176092e-07,
|
| 1558 |
+
"logits/chosen": 0.2258315533399582,
|
| 1559 |
+
"logits/rejected": 0.24323305487632751,
|
| 1560 |
+
"logps/chosen": -289.4067077636719,
|
| 1561 |
+
"logps/rejected": -317.7018737792969,
|
| 1562 |
+
"loss": 0.0448,
|
| 1563 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 1564 |
+
"rewards/chosen": -0.04625866562128067,
|
| 1565 |
+
"rewards/margins": 0.02912392094731331,
|
| 1566 |
+
"rewards/rejected": -0.07538258284330368,
|
| 1567 |
+
"step": 1000
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 0.79,
|
| 1571 |
+
"eval_logits/chosen": 0.21414244174957275,
|
| 1572 |
+
"eval_logits/rejected": 0.2463187277317047,
|
| 1573 |
+
"eval_logps/chosen": -324.4480285644531,
|
| 1574 |
+
"eval_logps/rejected": -335.3596496582031,
|
| 1575 |
+
"eval_loss": 0.040262602269649506,
|
| 1576 |
+
"eval_rewards/accuracies": 0.60628741979599,
|
| 1577 |
+
"eval_rewards/chosen": -0.054479535669088364,
|
| 1578 |
+
"eval_rewards/margins": 0.03244137018918991,
|
| 1579 |
+
"eval_rewards/rejected": -0.08692090213298798,
|
| 1580 |
+
"eval_runtime": 374.7358,
|
| 1581 |
+
"eval_samples_per_second": 5.337,
|
| 1582 |
+
"eval_steps_per_second": 0.446,
|
| 1583 |
+
"step": 1000
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 0.79,
|
| 1587 |
+
"learning_rate": 6.231355262852529e-07,
|
| 1588 |
+
"logits/chosen": 0.25781863927841187,
|
| 1589 |
+
"logits/rejected": 0.2564670443534851,
|
| 1590 |
+
"logps/chosen": -372.15350341796875,
|
| 1591 |
+
"logps/rejected": -347.10711669921875,
|
| 1592 |
+
"loss": 0.043,
|
| 1593 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1594 |
+
"rewards/chosen": -0.05305807664990425,
|
| 1595 |
+
"rewards/margins": 0.025652442127466202,
|
| 1596 |
+
"rewards/rejected": -0.07871051132678986,
|
| 1597 |
+
"step": 1010
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 0.8,
|
| 1601 |
+
"learning_rate": 5.785352205971275e-07,
|
| 1602 |
+
"logits/chosen": 0.19803152978420258,
|
| 1603 |
+
"logits/rejected": 0.23860549926757812,
|
| 1604 |
+
"logps/chosen": -348.16656494140625,
|
| 1605 |
+
"logps/rejected": -346.49188232421875,
|
| 1606 |
+
"loss": 0.0446,
|
| 1607 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1608 |
+
"rewards/chosen": -0.046660203486680984,
|
| 1609 |
+
"rewards/margins": 0.04340515658259392,
|
| 1610 |
+
"rewards/rejected": -0.0900653675198555,
|
| 1611 |
+
"step": 1020
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.81,
|
| 1615 |
+
"learning_rate": 5.353813333653287e-07,
|
| 1616 |
+
"logits/chosen": 0.18666525185108185,
|
| 1617 |
+
"logits/rejected": 0.2377820461988449,
|
| 1618 |
+
"logps/chosen": -311.5859680175781,
|
| 1619 |
+
"logps/rejected": -325.674072265625,
|
| 1620 |
+
"loss": 0.0346,
|
| 1621 |
+
"rewards/accuracies": 0.65625,
|
| 1622 |
+
"rewards/chosen": -0.04294878989458084,
|
| 1623 |
+
"rewards/margins": 0.04405210167169571,
|
| 1624 |
+
"rewards/rejected": -0.08700089901685715,
|
| 1625 |
+
"step": 1030
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 0.82,
|
| 1629 |
+
"learning_rate": 4.937063494834774e-07,
|
| 1630 |
+
"logits/chosen": 0.21636883914470673,
|
| 1631 |
+
"logits/rejected": 0.1967061161994934,
|
| 1632 |
+
"logps/chosen": -352.9900207519531,
|
| 1633 |
+
"logps/rejected": -355.19720458984375,
|
| 1634 |
+
"loss": 0.0415,
|
| 1635 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1636 |
+
"rewards/chosen": -0.052512962371110916,
|
| 1637 |
+
"rewards/margins": 0.03489057347178459,
|
| 1638 |
+
"rewards/rejected": -0.0874035432934761,
|
| 1639 |
+
"step": 1040
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 0.82,
|
| 1643 |
+
"learning_rate": 4.5354164057310857e-07,
|
| 1644 |
+
"logits/chosen": 0.19166240096092224,
|
| 1645 |
+
"logits/rejected": 0.2557406723499298,
|
| 1646 |
+
"logps/chosen": -322.88140869140625,
|
| 1647 |
+
"logps/rejected": -345.00189208984375,
|
| 1648 |
+
"loss": 0.03,
|
| 1649 |
+
"rewards/accuracies": 0.65625,
|
| 1650 |
+
"rewards/chosen": -0.05283380672335625,
|
| 1651 |
+
"rewards/margins": 0.04377451539039612,
|
| 1652 |
+
"rewards/rejected": -0.09660832583904266,
|
| 1653 |
+
"step": 1050
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.83,
|
| 1657 |
+
"learning_rate": 4.1491744136810066e-07,
|
| 1658 |
+
"logits/chosen": 0.18374313414096832,
|
| 1659 |
+
"logits/rejected": 0.24601753056049347,
|
| 1660 |
+
"logps/chosen": -308.51995849609375,
|
| 1661 |
+
"logps/rejected": -320.12347412109375,
|
| 1662 |
+
"loss": 0.0355,
|
| 1663 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1664 |
+
"rewards/chosen": -0.04724350571632385,
|
| 1665 |
+
"rewards/margins": 0.03354518115520477,
|
| 1666 |
+
"rewards/rejected": -0.08078868687152863,
|
| 1667 |
+
"step": 1060
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 0.84,
|
| 1671 |
+
"learning_rate": 3.7786282695491313e-07,
|
| 1672 |
+
"logits/chosen": 0.21392662823200226,
|
| 1673 |
+
"logits/rejected": 0.2224198579788208,
|
| 1674 |
+
"logps/chosen": -315.51959228515625,
|
| 1675 |
+
"logps/rejected": -305.6190490722656,
|
| 1676 |
+
"loss": 0.0421,
|
| 1677 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1678 |
+
"rewards/chosen": -0.03758303076028824,
|
| 1679 |
+
"rewards/margins": 0.04133836179971695,
|
| 1680 |
+
"rewards/rejected": -0.07892139256000519,
|
| 1681 |
+
"step": 1070
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 0.85,
|
| 1685 |
+
"learning_rate": 3.4240569088577564e-07,
|
| 1686 |
+
"logits/chosen": 0.2256053388118744,
|
| 1687 |
+
"logits/rejected": 0.23919573426246643,
|
| 1688 |
+
"logps/chosen": -322.1871337890625,
|
| 1689 |
+
"logps/rejected": -313.2369079589844,
|
| 1690 |
+
"loss": 0.0389,
|
| 1691 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1692 |
+
"rewards/chosen": -0.048378705978393555,
|
| 1693 |
+
"rewards/margins": 0.029405321925878525,
|
| 1694 |
+
"rewards/rejected": -0.07778403162956238,
|
| 1695 |
+
"step": 1080
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 0.86,
|
| 1699 |
+
"learning_rate": 3.0857272418129136e-07,
|
| 1700 |
+
"logits/chosen": 0.1939636766910553,
|
| 1701 |
+
"logits/rejected": 0.22864237427711487,
|
| 1702 |
+
"logps/chosen": -305.79156494140625,
|
| 1703 |
+
"logps/rejected": -313.49688720703125,
|
| 1704 |
+
"loss": 0.0375,
|
| 1705 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 1706 |
+
"rewards/chosen": -0.04112850874662399,
|
| 1707 |
+
"rewards/margins": 0.034687552601099014,
|
| 1708 |
+
"rewards/rejected": -0.07581605762243271,
|
| 1709 |
+
"step": 1090
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 0.86,
|
| 1713 |
+
"learning_rate": 2.7638939523827956e-07,
|
| 1714 |
+
"logits/chosen": 0.2385842353105545,
|
| 1715 |
+
"logits/rejected": 0.24533557891845703,
|
| 1716 |
+
"logps/chosen": -291.5062561035156,
|
| 1717 |
+
"logps/rejected": -320.8490295410156,
|
| 1718 |
+
"loss": 0.0411,
|
| 1719 |
+
"rewards/accuracies": 0.65625,
|
| 1720 |
+
"rewards/chosen": -0.04673101380467415,
|
| 1721 |
+
"rewards/margins": 0.043406564742326736,
|
| 1722 |
+
"rewards/rejected": -0.09013758599758148,
|
| 1723 |
+
"step": 1100
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 0.86,
|
| 1727 |
+
"eval_logits/chosen": 0.197061225771904,
|
| 1728 |
+
"eval_logits/rejected": 0.22829273343086243,
|
| 1729 |
+
"eval_logps/chosen": -324.82574462890625,
|
| 1730 |
+
"eval_logps/rejected": -336.86572265625,
|
| 1731 |
+
"eval_loss": 0.040152475237846375,
|
| 1732 |
+
"eval_rewards/accuracies": 0.6017963886260986,
|
| 1733 |
+
"eval_rewards/chosen": -0.05485724285244942,
|
| 1734 |
+
"eval_rewards/margins": 0.033569734543561935,
|
| 1735 |
+
"eval_rewards/rejected": -0.08842697739601135,
|
| 1736 |
+
"eval_runtime": 374.6149,
|
| 1737 |
+
"eval_samples_per_second": 5.339,
|
| 1738 |
+
"eval_steps_per_second": 0.446,
|
| 1739 |
+
"step": 1100
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.87,
|
| 1743 |
+
"learning_rate": 2.4587993065795983e-07,
|
| 1744 |
+
"logits/chosen": 0.23787248134613037,
|
| 1745 |
+
"logits/rejected": 0.25482678413391113,
|
| 1746 |
+
"logps/chosen": -336.42578125,
|
| 1747 |
+
"logps/rejected": -350.364990234375,
|
| 1748 |
+
"loss": 0.0449,
|
| 1749 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1750 |
+
"rewards/chosen": -0.047406163066625595,
|
| 1751 |
+
"rewards/margins": 0.04243654012680054,
|
| 1752 |
+
"rewards/rejected": -0.08984269946813583,
|
| 1753 |
+
"step": 1110
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.88,
|
| 1757 |
+
"learning_rate": 2.170672970089291e-07,
|
| 1758 |
+
"logits/chosen": 0.21486134827136993,
|
| 1759 |
+
"logits/rejected": 0.1970510333776474,
|
| 1760 |
+
"logps/chosen": -346.44989013671875,
|
| 1761 |
+
"logps/rejected": -328.222412109375,
|
| 1762 |
+
"loss": 0.0414,
|
| 1763 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 1764 |
+
"rewards/chosen": -0.05092350393533707,
|
| 1765 |
+
"rewards/margins": 0.022183910012245178,
|
| 1766 |
+
"rewards/rejected": -0.07310740649700165,
|
| 1767 |
+
"step": 1120
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.89,
|
| 1771 |
+
"learning_rate": 1.8997318353864673e-07,
|
| 1772 |
+
"logits/chosen": 0.18899409472942352,
|
| 1773 |
+
"logits/rejected": 0.21485233306884766,
|
| 1774 |
+
"logps/chosen": -315.77069091796875,
|
| 1775 |
+
"logps/rejected": -317.7997741699219,
|
| 1776 |
+
"loss": 0.0418,
|
| 1777 |
+
"rewards/accuracies": 0.65625,
|
| 1778 |
+
"rewards/chosen": -0.044594474136829376,
|
| 1779 |
+
"rewards/margins": 0.03823274374008179,
|
| 1780 |
+
"rewards/rejected": -0.08282722532749176,
|
| 1781 |
+
"step": 1130
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.89,
|
| 1785 |
+
"learning_rate": 1.6461798584644944e-07,
|
| 1786 |
+
"logits/chosen": 0.21227781474590302,
|
| 1787 |
+
"logits/rejected": 0.24548833072185516,
|
| 1788 |
+
"logps/chosen": -336.7071228027344,
|
| 1789 |
+
"logps/rejected": -352.16741943359375,
|
| 1790 |
+
"loss": 0.042,
|
| 1791 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1792 |
+
"rewards/chosen": -0.056326769292354584,
|
| 1793 |
+
"rewards/margins": 0.03633355721831322,
|
| 1794 |
+
"rewards/rejected": -0.0926603227853775,
|
| 1795 |
+
"step": 1140
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.9,
|
| 1799 |
+
"learning_rate": 1.4102079053038454e-07,
|
| 1800 |
+
"logits/chosen": 0.22621779143810272,
|
| 1801 |
+
"logits/rejected": 0.19496549665927887,
|
| 1802 |
+
"logps/chosen": -321.82904052734375,
|
| 1803 |
+
"logps/rejected": -300.1529846191406,
|
| 1804 |
+
"loss": 0.0416,
|
| 1805 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 1806 |
+
"rewards/chosen": -0.055761467665433884,
|
| 1807 |
+
"rewards/margins": 0.024841610342264175,
|
| 1808 |
+
"rewards/rejected": -0.08060307800769806,
|
| 1809 |
+
"step": 1150
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.91,
|
| 1813 |
+
"learning_rate": 1.1919936081941585e-07,
|
| 1814 |
+
"logits/chosen": 0.20498497784137726,
|
| 1815 |
+
"logits/rejected": 0.2435878962278366,
|
| 1816 |
+
"logps/chosen": -287.63934326171875,
|
| 1817 |
+
"logps/rejected": -335.10400390625,
|
| 1818 |
+
"loss": 0.039,
|
| 1819 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1820 |
+
"rewards/chosen": -0.0522816963493824,
|
| 1821 |
+
"rewards/margins": 0.04128851369023323,
|
| 1822 |
+
"rewards/rejected": -0.09357021003961563,
|
| 1823 |
+
"step": 1160
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 0.92,
|
| 1827 |
+
"learning_rate": 9.917012320182245e-08,
|
| 1828 |
+
"logits/chosen": 0.18155111372470856,
|
| 1829 |
+
"logits/rejected": 0.2222495973110199,
|
| 1830 |
+
"logps/chosen": -305.6929626464844,
|
| 1831 |
+
"logps/rejected": -338.02276611328125,
|
| 1832 |
+
"loss": 0.0464,
|
| 1833 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1834 |
+
"rewards/chosen": -0.04758179932832718,
|
| 1835 |
+
"rewards/margins": 0.03497401624917984,
|
| 1836 |
+
"rewards/rejected": -0.08255580812692642,
|
| 1837 |
+
"step": 1170
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.93,
|
| 1841 |
+
"learning_rate": 8.094815505985315e-08,
|
| 1842 |
+
"logits/chosen": 0.2116193324327469,
|
| 1843 |
+
"logits/rejected": 0.21422095596790314,
|
| 1844 |
+
"logps/chosen": -321.56341552734375,
|
| 1845 |
+
"logps/rejected": -308.92572021484375,
|
| 1846 |
+
"loss": 0.0384,
|
| 1847 |
+
"rewards/accuracies": 0.53125,
|
| 1848 |
+
"rewards/chosen": -0.05282514542341232,
|
| 1849 |
+
"rewards/margins": 0.021240225061774254,
|
| 1850 |
+
"rewards/rejected": -0.07406537234783173,
|
| 1851 |
+
"step": 1180
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.93,
|
| 1855 |
+
"learning_rate": 6.454717331994542e-08,
|
| 1856 |
+
"logits/chosen": 0.18755567073822021,
|
| 1857 |
+
"logits/rejected": 0.203456848859787,
|
| 1858 |
+
"logps/chosen": -321.9530334472656,
|
| 1859 |
+
"logps/rejected": -328.56524658203125,
|
| 1860 |
+
"loss": 0.0416,
|
| 1861 |
+
"rewards/accuracies": 0.59375,
|
| 1862 |
+
"rewards/chosen": -0.04693279415369034,
|
| 1863 |
+
"rewards/margins": 0.028221463784575462,
|
| 1864 |
+
"rewards/rejected": -0.07515425980091095,
|
| 1865 |
+
"step": 1190
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 0.94,
|
| 1869 |
+
"learning_rate": 4.9979524127052595e-08,
|
| 1870 |
+
"logits/chosen": 0.20014750957489014,
|
| 1871 |
+
"logits/rejected": 0.21643492579460144,
|
| 1872 |
+
"logps/chosen": -325.4107666015625,
|
| 1873 |
+
"logps/rejected": -321.22515869140625,
|
| 1874 |
+
"loss": 0.0459,
|
| 1875 |
+
"rewards/accuracies": 0.543749988079071,
|
| 1876 |
+
"rewards/chosen": -0.04905826970934868,
|
| 1877 |
+
"rewards/margins": 0.03321439027786255,
|
| 1878 |
+
"rewards/rejected": -0.08227266371250153,
|
| 1879 |
+
"step": 1200
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.94,
|
| 1883 |
+
"eval_logits/chosen": 0.18975894153118134,
|
| 1884 |
+
"eval_logits/rejected": 0.22052054107189178,
|
| 1885 |
+
"eval_logps/chosen": -325.7256774902344,
|
| 1886 |
+
"eval_logps/rejected": -338.00421142578125,
|
| 1887 |
+
"eval_loss": 0.04012565314769745,
|
| 1888 |
+
"eval_rewards/accuracies": 0.6032934188842773,
|
| 1889 |
+
"eval_rewards/chosen": -0.05575716868042946,
|
| 1890 |
+
"eval_rewards/margins": 0.03380833566188812,
|
| 1891 |
+
"eval_rewards/rejected": -0.08956550806760788,
|
| 1892 |
+
"eval_runtime": 374.5923,
|
| 1893 |
+
"eval_samples_per_second": 5.339,
|
| 1894 |
+
"eval_steps_per_second": 0.446,
|
| 1895 |
+
"step": 1200
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 0.95,
|
| 1899 |
+
"learning_rate": 3.725617355085476e-08,
|
| 1900 |
+
"logits/chosen": 0.22028300166130066,
|
| 1901 |
+
"logits/rejected": 0.26091286540031433,
|
| 1902 |
+
"logps/chosen": -323.34808349609375,
|
| 1903 |
+
"logps/rejected": -322.25201416015625,
|
| 1904 |
+
"loss": 0.038,
|
| 1905 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1906 |
+
"rewards/chosen": -0.049160152673721313,
|
| 1907 |
+
"rewards/margins": 0.03982594236731529,
|
| 1908 |
+
"rewards/rejected": -0.08898608386516571,
|
| 1909 |
+
"step": 1210
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 0.96,
|
| 1913 |
+
"learning_rate": 2.63866993308437e-08,
|
| 1914 |
+
"logits/chosen": 0.21131488680839539,
|
| 1915 |
+
"logits/rejected": 0.25622323155403137,
|
| 1916 |
+
"logps/chosen": -347.6703796386719,
|
| 1917 |
+
"logps/rejected": -305.52484130859375,
|
| 1918 |
+
"loss": 0.0427,
|
| 1919 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1920 |
+
"rewards/chosen": -0.04987090080976486,
|
| 1921 |
+
"rewards/margins": 0.02888522483408451,
|
| 1922 |
+
"rewards/rejected": -0.07875612378120422,
|
| 1923 |
+
"step": 1220
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 0.97,
|
| 1927 |
+
"learning_rate": 1.737928366650099e-08,
|
| 1928 |
+
"logits/chosen": 0.1875266432762146,
|
| 1929 |
+
"logits/rejected": 0.23892006278038025,
|
| 1930 |
+
"logps/chosen": -362.18450927734375,
|
| 1931 |
+
"logps/rejected": -347.82489013671875,
|
| 1932 |
+
"loss": 0.0424,
|
| 1933 |
+
"rewards/accuracies": 0.65625,
|
| 1934 |
+
"rewards/chosen": -0.05079431086778641,
|
| 1935 |
+
"rewards/margins": 0.03944983333349228,
|
| 1936 |
+
"rewards/rejected": -0.09024415165185928,
|
| 1937 |
+
"step": 1230
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 0.97,
|
| 1941 |
+
"learning_rate": 1.0240707057995735e-08,
|
| 1942 |
+
"logits/chosen": 0.1739097535610199,
|
| 1943 |
+
"logits/rejected": 0.18980395793914795,
|
| 1944 |
+
"logps/chosen": -297.0783996582031,
|
| 1945 |
+
"logps/rejected": -328.2581787109375,
|
| 1946 |
+
"loss": 0.039,
|
| 1947 |
+
"rewards/accuracies": 0.675000011920929,
|
| 1948 |
+
"rewards/chosen": -0.04551629722118378,
|
| 1949 |
+
"rewards/margins": 0.04395337030291557,
|
| 1950 |
+
"rewards/rejected": -0.08946967869997025,
|
| 1951 |
+
"step": 1240
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 0.98,
|
| 1955 |
+
"learning_rate": 4.976343202034717e-09,
|
| 1956 |
+
"logits/chosen": 0.20210778713226318,
|
| 1957 |
+
"logits/rejected": 0.21424946188926697,
|
| 1958 |
+
"logps/chosen": -314.2582092285156,
|
| 1959 |
+
"logps/rejected": -324.7920837402344,
|
| 1960 |
+
"loss": 0.0368,
|
| 1961 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1962 |
+
"rewards/chosen": -0.048636116087436676,
|
| 1963 |
+
"rewards/margins": 0.03599988669157028,
|
| 1964 |
+
"rewards/rejected": -0.08463600277900696,
|
| 1965 |
+
"step": 1250
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.99,
|
| 1969 |
+
"learning_rate": 1.5901549467139953e-09,
|
| 1970 |
+
"logits/chosen": 0.2138410359621048,
|
| 1971 |
+
"logits/rejected": 0.24039845168590546,
|
| 1972 |
+
"logps/chosen": -289.2809143066406,
|
| 1973 |
+
"logps/rejected": -290.346923828125,
|
| 1974 |
+
"loss": 0.0383,
|
| 1975 |
+
"rewards/accuracies": 0.625,
|
| 1976 |
+
"rewards/chosen": -0.045187849551439285,
|
| 1977 |
+
"rewards/margins": 0.036808837205171585,
|
| 1978 |
+
"rewards/rejected": -0.08199669420719147,
|
| 1979 |
+
"step": 1260
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 1.0,
|
| 1983 |
+
"learning_rate": 8.469130840960127e-11,
|
| 1984 |
+
"logits/chosen": 0.22721895575523376,
|
| 1985 |
+
"logits/rejected": 0.24769964814186096,
|
| 1986 |
+
"logps/chosen": -305.83270263671875,
|
| 1987 |
+
"logps/rejected": -334.0370788574219,
|
| 1988 |
+
"loss": 0.0414,
|
| 1989 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1990 |
+
"rewards/chosen": -0.048570211976766586,
|
| 1991 |
+
"rewards/margins": 0.0430789478123188,
|
| 1992 |
+
"rewards/rejected": -0.09164915978908539,
|
| 1993 |
+
"step": 1270
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 1.0,
|
| 1997 |
+
"step": 1273,
|
| 1998 |
+
"total_flos": 0.0,
|
| 1999 |
+
"train_loss": 0.044965725131654775,
|
| 2000 |
+
"train_runtime": 22244.9719,
|
| 2001 |
+
"train_samples_per_second": 2.748,
|
| 2002 |
+
"train_steps_per_second": 0.057
|
| 2003 |
+
}
|
| 2004 |
+
],
|
| 2005 |
+
"logging_steps": 10,
|
| 2006 |
+
"max_steps": 1273,
|
| 2007 |
+
"num_input_tokens_seen": 0,
|
| 2008 |
+
"num_train_epochs": 1,
|
| 2009 |
+
"save_steps": 100,
|
| 2010 |
+
"total_flos": 0.0,
|
| 2011 |
+
"train_batch_size": 4,
|
| 2012 |
+
"trial_name": null,
|
| 2013 |
+
"trial_params": null
|
| 2014 |
+
}
|