Model save
Browse files- README.md +82 -0
- all_results.json +21 -0
- eval_results.json +16 -0
- train_results.json +8 -0
- trainer_state.json +1604 -0
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-gpo-iter1
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-7b-gpo-iter1
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.0069
|
| 22 |
+
- Rewards/chosen: 0.0025
|
| 23 |
+
- Rewards/rejected: 0.0081
|
| 24 |
+
- Rewards/accuracies: 0.4595
|
| 25 |
+
- Rewards/margins: -0.0056
|
| 26 |
+
- Logps/rejected: -272.5866
|
| 27 |
+
- Logps/chosen: -298.8498
|
| 28 |
+
- Logits/rejected: -2.1749
|
| 29 |
+
- Logits/chosen: -2.3692
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-06
|
| 49 |
+
- train_batch_size: 1
|
| 50 |
+
- eval_batch_size: 2
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- gradient_accumulation_steps: 2
|
| 54 |
+
- total_train_batch_size: 2
|
| 55 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 56 |
+
- lr_scheduler_type: cosine
|
| 57 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 58 |
+
- num_epochs: 2
|
| 59 |
+
|
| 60 |
+
### Training results
|
| 61 |
+
|
| 62 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 63 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 64 |
+
| 0.0006 | 0.2 | 100 | 0.0031 | -0.0541 | -0.0467 | 0.4245 | -0.0074 | -278.0669 | -304.5065 | -2.1506 | -2.3436 |
|
| 65 |
+
| 0.0025 | 0.4 | 200 | 0.0033 | -0.0115 | -0.0107 | 0.4910 | -0.0008 | -274.4619 | -300.2420 | -2.1684 | -2.3612 |
|
| 66 |
+
| 0.0009 | 0.6 | 300 | 0.0030 | -0.0220 | -0.0216 | 0.4935 | -0.0004 | -275.5567 | -301.2960 | -2.1427 | -2.3360 |
|
| 67 |
+
| 0.0013 | 0.8 | 400 | 0.0034 | -0.0156 | -0.0142 | 0.4935 | -0.0014 | -274.8156 | -300.6561 | -2.1462 | -2.3405 |
|
| 68 |
+
| 0.0011 | 1.0 | 500 | 0.0037 | -0.0565 | -0.0502 | 0.4520 | -0.0063 | -278.4165 | -304.7457 | -2.1454 | -2.3392 |
|
| 69 |
+
| 0.0116 | 1.2 | 600 | 0.0049 | -0.0283 | -0.0229 | 0.4435 | -0.0054 | -275.6791 | -301.9266 | -2.1527 | -2.3449 |
|
| 70 |
+
| 0.015 | 1.4 | 700 | 0.0065 | -0.0261 | -0.0182 | 0.4450 | -0.0078 | -275.2170 | -301.7041 | -2.1650 | -2.3586 |
|
| 71 |
+
| 0.0009 | 1.6 | 800 | 0.0069 | 0.0079 | 0.0124 | 0.4720 | -0.0044 | -272.1540 | -298.3011 | -2.1746 | -2.3689 |
|
| 72 |
+
| 0.0109 | 1.8 | 900 | 0.0069 | 0.0024 | 0.0080 | 0.4570 | -0.0057 | -272.5880 | -298.8583 | -2.1739 | -2.3682 |
|
| 73 |
+
| 0.0015 | 2.0 | 1000 | 0.0069 | 0.0025 | 0.0081 | 0.4595 | -0.0056 | -272.5866 | -298.8498 | -2.1749 | -2.3692 |
|
| 74 |
+
|
| 75 |
+
|
| 76 |
+
### Framework versions
|
| 77 |
+
|
| 78 |
+
- PEFT 0.7.1
|
| 79 |
+
- Transformers 4.36.2
|
| 80 |
+
- Pytorch 2.1.2+cu118
|
| 81 |
+
- Datasets 2.14.6
|
| 82 |
+
- Tokenizers 0.15.2
|
all_results.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_logits/chosen": -2.3691599369049072,
|
| 4 |
+
"eval_logits/rejected": -2.1749157905578613,
|
| 5 |
+
"eval_logps/chosen": -298.8498229980469,
|
| 6 |
+
"eval_logps/rejected": -272.58660888671875,
|
| 7 |
+
"eval_loss": 0.006900906562805176,
|
| 8 |
+
"eval_rewards/accuracies": 0.4595000147819519,
|
| 9 |
+
"eval_rewards/chosen": 0.002455136040225625,
|
| 10 |
+
"eval_rewards/margins": -0.005604185629636049,
|
| 11 |
+
"eval_rewards/rejected": 0.008059320971369743,
|
| 12 |
+
"eval_runtime": 1421.1253,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 1.407,
|
| 15 |
+
"eval_steps_per_second": 0.704,
|
| 16 |
+
"train_loss": 0.005910432943725027,
|
| 17 |
+
"train_runtime": 19053.6119,
|
| 18 |
+
"train_samples": 61135,
|
| 19 |
+
"train_samples_per_second": 0.105,
|
| 20 |
+
"train_steps_per_second": 0.052
|
| 21 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_logits/chosen": -2.3691599369049072,
|
| 4 |
+
"eval_logits/rejected": -2.1749157905578613,
|
| 5 |
+
"eval_logps/chosen": -298.8498229980469,
|
| 6 |
+
"eval_logps/rejected": -272.58660888671875,
|
| 7 |
+
"eval_loss": 0.006900906562805176,
|
| 8 |
+
"eval_rewards/accuracies": 0.4595000147819519,
|
| 9 |
+
"eval_rewards/chosen": 0.002455136040225625,
|
| 10 |
+
"eval_rewards/margins": -0.005604185629636049,
|
| 11 |
+
"eval_rewards/rejected": 0.008059320971369743,
|
| 12 |
+
"eval_runtime": 1421.1253,
|
| 13 |
+
"eval_samples": 2000,
|
| 14 |
+
"eval_samples_per_second": 1.407,
|
| 15 |
+
"eval_steps_per_second": 0.704
|
| 16 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.005910432943725027,
|
| 4 |
+
"train_runtime": 19053.6119,
|
| 5 |
+
"train_samples": 61135,
|
| 6 |
+
"train_samples_per_second": 0.105,
|
| 7 |
+
"train_steps_per_second": 0.052
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1604 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1000,
|
| 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": 5.0000000000000004e-08,
|
| 14 |
+
"logits/chosen": -3.1678528785705566,
|
| 15 |
+
"logits/rejected": -3.1461400985717773,
|
| 16 |
+
"logps/chosen": -344.33587646484375,
|
| 17 |
+
"logps/rejected": -334.830810546875,
|
| 18 |
+
"loss": 0.0011,
|
| 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": 0.02,
|
| 27 |
+
"learning_rate": 5.000000000000001e-07,
|
| 28 |
+
"logits/chosen": -2.8056228160858154,
|
| 29 |
+
"logits/rejected": -2.711918830871582,
|
| 30 |
+
"logps/chosen": -137.3001708984375,
|
| 31 |
+
"logps/rejected": -114.56926727294922,
|
| 32 |
+
"loss": 0.001,
|
| 33 |
+
"rewards/accuracies": 0.6111111044883728,
|
| 34 |
+
"rewards/chosen": 0.0003753364726435393,
|
| 35 |
+
"rewards/margins": 0.001190218492411077,
|
| 36 |
+
"rewards/rejected": -0.0008148818160407245,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04,
|
| 41 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 42 |
+
"logits/chosen": -2.664208173751831,
|
| 43 |
+
"logits/rejected": -2.3068690299987793,
|
| 44 |
+
"logps/chosen": -95.76632690429688,
|
| 45 |
+
"logps/rejected": -91.30467224121094,
|
| 46 |
+
"loss": 0.0011,
|
| 47 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 48 |
+
"rewards/chosen": -7.150838064262643e-05,
|
| 49 |
+
"rewards/margins": 0.00038482382660731673,
|
| 50 |
+
"rewards/rejected": -0.0004563321708701551,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.06,
|
| 55 |
+
"learning_rate": 1.5e-06,
|
| 56 |
+
"logits/chosen": -2.8235249519348145,
|
| 57 |
+
"logits/rejected": -2.742885112762451,
|
| 58 |
+
"logps/chosen": -103.0609359741211,
|
| 59 |
+
"logps/rejected": -119.29620361328125,
|
| 60 |
+
"loss": 0.0011,
|
| 61 |
+
"rewards/accuracies": 0.550000011920929,
|
| 62 |
+
"rewards/chosen": -0.00019755656830966473,
|
| 63 |
+
"rewards/margins": 0.00018337149231228977,
|
| 64 |
+
"rewards/rejected": -0.00038092798786237836,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08,
|
| 69 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 70 |
+
"logits/chosen": -2.812087059020996,
|
| 71 |
+
"logits/rejected": -2.3899402618408203,
|
| 72 |
+
"logps/chosen": -56.24201583862305,
|
| 73 |
+
"logps/rejected": -52.55266189575195,
|
| 74 |
+
"loss": 0.001,
|
| 75 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 76 |
+
"rewards/chosen": -0.002384146675467491,
|
| 77 |
+
"rewards/margins": 0.0011411935556679964,
|
| 78 |
+
"rewards/rejected": -0.0035253397654742002,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.1,
|
| 83 |
+
"learning_rate": 2.5e-06,
|
| 84 |
+
"logits/chosen": -2.919544219970703,
|
| 85 |
+
"logits/rejected": -2.68827486038208,
|
| 86 |
+
"logps/chosen": -85.77859497070312,
|
| 87 |
+
"logps/rejected": -82.93376159667969,
|
| 88 |
+
"loss": 0.001,
|
| 89 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 90 |
+
"rewards/chosen": -0.002311713993549347,
|
| 91 |
+
"rewards/margins": 0.0031323134899139404,
|
| 92 |
+
"rewards/rejected": -0.005444027483463287,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.12,
|
| 97 |
+
"learning_rate": 3e-06,
|
| 98 |
+
"logits/chosen": -2.7442586421966553,
|
| 99 |
+
"logits/rejected": -2.5409207344055176,
|
| 100 |
+
"logps/chosen": -135.08984375,
|
| 101 |
+
"logps/rejected": -137.5052032470703,
|
| 102 |
+
"loss": 0.001,
|
| 103 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 104 |
+
"rewards/chosen": -0.0012713891919702291,
|
| 105 |
+
"rewards/margins": 0.005557497497648001,
|
| 106 |
+
"rewards/rejected": -0.006828887853771448,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.14,
|
| 111 |
+
"learning_rate": 3.5e-06,
|
| 112 |
+
"logits/chosen": -2.680795669555664,
|
| 113 |
+
"logits/rejected": -2.3159432411193848,
|
| 114 |
+
"logps/chosen": -86.95841979980469,
|
| 115 |
+
"logps/rejected": -71.47272491455078,
|
| 116 |
+
"loss": 0.0012,
|
| 117 |
+
"rewards/accuracies": 0.550000011920929,
|
| 118 |
+
"rewards/chosen": 0.0019758909475058317,
|
| 119 |
+
"rewards/margins": 0.005556103773415089,
|
| 120 |
+
"rewards/rejected": -0.0035802137572318316,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.16,
|
| 125 |
+
"learning_rate": 4.000000000000001e-06,
|
| 126 |
+
"logits/chosen": -2.789736270904541,
|
| 127 |
+
"logits/rejected": -2.5091891288757324,
|
| 128 |
+
"logps/chosen": -54.42887496948242,
|
| 129 |
+
"logps/rejected": -45.48096466064453,
|
| 130 |
+
"loss": 0.0009,
|
| 131 |
+
"rewards/accuracies": 0.699999988079071,
|
| 132 |
+
"rewards/chosen": 0.011168469674885273,
|
| 133 |
+
"rewards/margins": 0.00967374723404646,
|
| 134 |
+
"rewards/rejected": 0.0014947212766855955,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.18,
|
| 139 |
+
"learning_rate": 4.5e-06,
|
| 140 |
+
"logits/chosen": -2.686554431915283,
|
| 141 |
+
"logits/rejected": -2.3966121673583984,
|
| 142 |
+
"logps/chosen": -88.09907531738281,
|
| 143 |
+
"logps/rejected": -87.56222534179688,
|
| 144 |
+
"loss": 0.0008,
|
| 145 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 146 |
+
"rewards/chosen": 0.016689006239175797,
|
| 147 |
+
"rewards/margins": 0.0082827378064394,
|
| 148 |
+
"rewards/rejected": 0.008406268432736397,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2,
|
| 153 |
+
"learning_rate": 5e-06,
|
| 154 |
+
"logits/chosen": -2.821899652481079,
|
| 155 |
+
"logits/rejected": -2.5310721397399902,
|
| 156 |
+
"logps/chosen": -64.77155303955078,
|
| 157 |
+
"logps/rejected": -73.11659240722656,
|
| 158 |
+
"loss": 0.0006,
|
| 159 |
+
"rewards/accuracies": 0.699999988079071,
|
| 160 |
+
"rewards/chosen": 0.018898095935583115,
|
| 161 |
+
"rewards/margins": 0.017503971233963966,
|
| 162 |
+
"rewards/rejected": 0.001394127495586872,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.2,
|
| 167 |
+
"eval_logits/chosen": -2.343594551086426,
|
| 168 |
+
"eval_logits/rejected": -2.15059494972229,
|
| 169 |
+
"eval_logps/chosen": -304.50653076171875,
|
| 170 |
+
"eval_logps/rejected": -278.0669250488281,
|
| 171 |
+
"eval_loss": 0.003129809396341443,
|
| 172 |
+
"eval_rewards/accuracies": 0.4244999885559082,
|
| 173 |
+
"eval_rewards/chosen": -0.054112162441015244,
|
| 174 |
+
"eval_rewards/margins": -0.007367968093603849,
|
| 175 |
+
"eval_rewards/rejected": -0.046744197607040405,
|
| 176 |
+
"eval_runtime": 1422.4177,
|
| 177 |
+
"eval_samples_per_second": 1.406,
|
| 178 |
+
"eval_steps_per_second": 0.703,
|
| 179 |
+
"step": 100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.22,
|
| 183 |
+
"learning_rate": 4.99847706754774e-06,
|
| 184 |
+
"logits/chosen": -2.8732924461364746,
|
| 185 |
+
"logits/rejected": -2.670675754547119,
|
| 186 |
+
"logps/chosen": -80.2041244506836,
|
| 187 |
+
"logps/rejected": -76.78239440917969,
|
| 188 |
+
"loss": 0.0007,
|
| 189 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 190 |
+
"rewards/chosen": 0.0098586305975914,
|
| 191 |
+
"rewards/margins": 0.01424575038254261,
|
| 192 |
+
"rewards/rejected": -0.004387119319289923,
|
| 193 |
+
"step": 110
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.24,
|
| 197 |
+
"learning_rate": 4.993910125649561e-06,
|
| 198 |
+
"logits/chosen": -2.841202735900879,
|
| 199 |
+
"logits/rejected": -2.513766050338745,
|
| 200 |
+
"logps/chosen": -60.289215087890625,
|
| 201 |
+
"logps/rejected": -46.964500427246094,
|
| 202 |
+
"loss": 0.0011,
|
| 203 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 204 |
+
"rewards/chosen": 0.013731713406741619,
|
| 205 |
+
"rewards/margins": 0.013367941603064537,
|
| 206 |
+
"rewards/rejected": 0.00036377101787365973,
|
| 207 |
+
"step": 120
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.26,
|
| 211 |
+
"learning_rate": 4.986304738420684e-06,
|
| 212 |
+
"logits/chosen": -2.712897777557373,
|
| 213 |
+
"logits/rejected": -2.391780138015747,
|
| 214 |
+
"logps/chosen": -134.18972778320312,
|
| 215 |
+
"logps/rejected": -118.10213470458984,
|
| 216 |
+
"loss": 0.0011,
|
| 217 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 218 |
+
"rewards/chosen": 0.0306332316249609,
|
| 219 |
+
"rewards/margins": 0.015287762507796288,
|
| 220 |
+
"rewards/rejected": 0.015345467254519463,
|
| 221 |
+
"step": 130
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.28,
|
| 225 |
+
"learning_rate": 4.975670171853926e-06,
|
| 226 |
+
"logits/chosen": -2.9027531147003174,
|
| 227 |
+
"logits/rejected": -2.519896984100342,
|
| 228 |
+
"logps/chosen": -57.02116012573242,
|
| 229 |
+
"logps/rejected": -53.072364807128906,
|
| 230 |
+
"loss": 0.0009,
|
| 231 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 232 |
+
"rewards/chosen": -0.0006130338879302144,
|
| 233 |
+
"rewards/margins": 0.008738933131098747,
|
| 234 |
+
"rewards/rejected": -0.00935196690261364,
|
| 235 |
+
"step": 140
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.3,
|
| 239 |
+
"learning_rate": 4.962019382530521e-06,
|
| 240 |
+
"logits/chosen": -2.8638014793395996,
|
| 241 |
+
"logits/rejected": -2.5246753692626953,
|
| 242 |
+
"logps/chosen": -61.296531677246094,
|
| 243 |
+
"logps/rejected": -76.2929916381836,
|
| 244 |
+
"loss": 0.0014,
|
| 245 |
+
"rewards/accuracies": 0.699999988079071,
|
| 246 |
+
"rewards/chosen": -0.025341859087347984,
|
| 247 |
+
"rewards/margins": 0.015296203084290028,
|
| 248 |
+
"rewards/rejected": -0.04063806310296059,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.32,
|
| 253 |
+
"learning_rate": 4.9453690018345144e-06,
|
| 254 |
+
"logits/chosen": -2.8262181282043457,
|
| 255 |
+
"logits/rejected": -2.3856072425842285,
|
| 256 |
+
"logps/chosen": -141.7677459716797,
|
| 257 |
+
"logps/rejected": -134.89096069335938,
|
| 258 |
+
"loss": 0.001,
|
| 259 |
+
"rewards/accuracies": 0.699999988079071,
|
| 260 |
+
"rewards/chosen": 0.005657578818500042,
|
| 261 |
+
"rewards/margins": 0.017082063481211662,
|
| 262 |
+
"rewards/rejected": -0.011424483731389046,
|
| 263 |
+
"step": 160
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.34,
|
| 267 |
+
"learning_rate": 4.925739315689991e-06,
|
| 268 |
+
"logits/chosen": -2.871880531311035,
|
| 269 |
+
"logits/rejected": -2.5630600452423096,
|
| 270 |
+
"logps/chosen": -61.668121337890625,
|
| 271 |
+
"logps/rejected": -58.292213439941406,
|
| 272 |
+
"loss": 0.0011,
|
| 273 |
+
"rewards/accuracies": 0.699999988079071,
|
| 274 |
+
"rewards/chosen": -0.016033068299293518,
|
| 275 |
+
"rewards/margins": 0.01451338641345501,
|
| 276 |
+
"rewards/rejected": -0.030546456575393677,
|
| 277 |
+
"step": 170
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.36,
|
| 281 |
+
"learning_rate": 4.903154239845798e-06,
|
| 282 |
+
"logits/chosen": -2.893052339553833,
|
| 283 |
+
"logits/rejected": -2.682218074798584,
|
| 284 |
+
"logps/chosen": -92.70951080322266,
|
| 285 |
+
"logps/rejected": -104.5350570678711,
|
| 286 |
+
"loss": 0.0014,
|
| 287 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 288 |
+
"rewards/chosen": -0.004353048745542765,
|
| 289 |
+
"rewards/margins": 0.005879141390323639,
|
| 290 |
+
"rewards/rejected": -0.010232190601527691,
|
| 291 |
+
"step": 180
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.38,
|
| 295 |
+
"learning_rate": 4.8776412907378845e-06,
|
| 296 |
+
"logits/chosen": -2.6648871898651123,
|
| 297 |
+
"logits/rejected": -2.4408602714538574,
|
| 298 |
+
"logps/chosen": -68.06556701660156,
|
| 299 |
+
"logps/rejected": -83.47774505615234,
|
| 300 |
+
"loss": 0.0016,
|
| 301 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 302 |
+
"rewards/chosen": -0.023349497467279434,
|
| 303 |
+
"rewards/margins": 0.0038781266193836927,
|
| 304 |
+
"rewards/rejected": -0.027227621525526047,
|
| 305 |
+
"step": 190
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.4,
|
| 309 |
+
"learning_rate": 4.849231551964771e-06,
|
| 310 |
+
"logits/chosen": -2.9131691455841064,
|
| 311 |
+
"logits/rejected": -2.6140384674072266,
|
| 312 |
+
"logps/chosen": -118.0514907836914,
|
| 313 |
+
"logps/rejected": -105.72004699707031,
|
| 314 |
+
"loss": 0.0025,
|
| 315 |
+
"rewards/accuracies": 0.550000011920929,
|
| 316 |
+
"rewards/chosen": -0.01497165858745575,
|
| 317 |
+
"rewards/margins": 0.009273095987737179,
|
| 318 |
+
"rewards/rejected": -0.024244757369160652,
|
| 319 |
+
"step": 200
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.4,
|
| 323 |
+
"eval_logits/chosen": -2.3611958026885986,
|
| 324 |
+
"eval_logits/rejected": -2.168394088745117,
|
| 325 |
+
"eval_logps/chosen": -300.2420349121094,
|
| 326 |
+
"eval_logps/rejected": -274.46185302734375,
|
| 327 |
+
"eval_loss": 0.0032995175570249557,
|
| 328 |
+
"eval_rewards/accuracies": 0.4909999966621399,
|
| 329 |
+
"eval_rewards/chosen": -0.011466986499726772,
|
| 330 |
+
"eval_rewards/margins": -0.0007737927371636033,
|
| 331 |
+
"eval_rewards/rejected": -0.010693194344639778,
|
| 332 |
+
"eval_runtime": 1422.8869,
|
| 333 |
+
"eval_samples_per_second": 1.406,
|
| 334 |
+
"eval_steps_per_second": 0.703,
|
| 335 |
+
"step": 200
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.42,
|
| 339 |
+
"learning_rate": 4.817959636416969e-06,
|
| 340 |
+
"logits/chosen": -2.7499923706054688,
|
| 341 |
+
"logits/rejected": -2.4605581760406494,
|
| 342 |
+
"logps/chosen": -60.11127471923828,
|
| 343 |
+
"logps/rejected": -57.330078125,
|
| 344 |
+
"loss": 0.0008,
|
| 345 |
+
"rewards/accuracies": 0.75,
|
| 346 |
+
"rewards/chosen": 2.1754764020442963e-05,
|
| 347 |
+
"rewards/margins": 0.01557920128107071,
|
| 348 |
+
"rewards/rejected": -0.015557448379695415,
|
| 349 |
+
"step": 210
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.44,
|
| 353 |
+
"learning_rate": 4.783863644106502e-06,
|
| 354 |
+
"logits/chosen": -2.8885045051574707,
|
| 355 |
+
"logits/rejected": -2.744602680206299,
|
| 356 |
+
"logps/chosen": -133.10980224609375,
|
| 357 |
+
"logps/rejected": -116.32179260253906,
|
| 358 |
+
"loss": 0.0013,
|
| 359 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 360 |
+
"rewards/chosen": 0.012220713309943676,
|
| 361 |
+
"rewards/margins": 0.013802575878798962,
|
| 362 |
+
"rewards/rejected": -0.0015818632673472166,
|
| 363 |
+
"step": 220
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.46,
|
| 367 |
+
"learning_rate": 4.746985115747918e-06,
|
| 368 |
+
"logits/chosen": -2.6919429302215576,
|
| 369 |
+
"logits/rejected": -2.4415671825408936,
|
| 370 |
+
"logps/chosen": -57.36275100708008,
|
| 371 |
+
"logps/rejected": -43.19419479370117,
|
| 372 |
+
"loss": 0.0008,
|
| 373 |
+
"rewards/accuracies": 0.75,
|
| 374 |
+
"rewards/chosen": 0.01155207958072424,
|
| 375 |
+
"rewards/margins": 0.017826344817876816,
|
| 376 |
+
"rewards/rejected": -0.0062742652371525764,
|
| 377 |
+
"step": 230
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.48,
|
| 381 |
+
"learning_rate": 4.707368982147318e-06,
|
| 382 |
+
"logits/chosen": -2.7176296710968018,
|
| 383 |
+
"logits/rejected": -2.6274712085723877,
|
| 384 |
+
"logps/chosen": -59.256858825683594,
|
| 385 |
+
"logps/rejected": -60.37900924682617,
|
| 386 |
+
"loss": 0.0019,
|
| 387 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 388 |
+
"rewards/chosen": 0.004980463068932295,
|
| 389 |
+
"rewards/margins": 0.006475840695202351,
|
| 390 |
+
"rewards/rejected": -0.0014953784411773086,
|
| 391 |
+
"step": 240
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.5,
|
| 395 |
+
"learning_rate": 4.665063509461098e-06,
|
| 396 |
+
"logits/chosen": -2.8012423515319824,
|
| 397 |
+
"logits/rejected": -2.525057554244995,
|
| 398 |
+
"logps/chosen": -98.50082397460938,
|
| 399 |
+
"logps/rejected": -90.67588806152344,
|
| 400 |
+
"loss": 0.0013,
|
| 401 |
+
"rewards/accuracies": 0.699999988079071,
|
| 402 |
+
"rewards/chosen": -0.008941109292209148,
|
| 403 |
+
"rewards/margins": 0.012530329637229443,
|
| 404 |
+
"rewards/rejected": -0.021471437066793442,
|
| 405 |
+
"step": 250
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.52,
|
| 409 |
+
"learning_rate": 4.620120240391065e-06,
|
| 410 |
+
"logits/chosen": -2.762904644012451,
|
| 411 |
+
"logits/rejected": -2.579153537750244,
|
| 412 |
+
"logps/chosen": -79.9581069946289,
|
| 413 |
+
"logps/rejected": -84.54290771484375,
|
| 414 |
+
"loss": 0.0016,
|
| 415 |
+
"rewards/accuracies": 0.550000011920929,
|
| 416 |
+
"rewards/chosen": -0.008370987139642239,
|
| 417 |
+
"rewards/margins": 0.0011044790735468268,
|
| 418 |
+
"rewards/rejected": -0.009475464932620525,
|
| 419 |
+
"step": 260
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.54,
|
| 423 |
+
"learning_rate": 4.572593931387604e-06,
|
| 424 |
+
"logits/chosen": -2.697049140930176,
|
| 425 |
+
"logits/rejected": -2.420090675354004,
|
| 426 |
+
"logps/chosen": -68.5470199584961,
|
| 427 |
+
"logps/rejected": -59.51050567626953,
|
| 428 |
+
"loss": 0.0011,
|
| 429 |
+
"rewards/accuracies": 0.550000011920929,
|
| 430 |
+
"rewards/chosen": -0.007652081549167633,
|
| 431 |
+
"rewards/margins": 0.009131882339715958,
|
| 432 |
+
"rewards/rejected": -0.01678396575152874,
|
| 433 |
+
"step": 270
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.56,
|
| 437 |
+
"learning_rate": 4.522542485937369e-06,
|
| 438 |
+
"logits/chosen": -2.8917226791381836,
|
| 439 |
+
"logits/rejected": -2.56544828414917,
|
| 440 |
+
"logps/chosen": -133.83926391601562,
|
| 441 |
+
"logps/rejected": -106.41746520996094,
|
| 442 |
+
"loss": 0.0014,
|
| 443 |
+
"rewards/accuracies": 0.4000000059604645,
|
| 444 |
+
"rewards/chosen": 0.0028765772003680468,
|
| 445 |
+
"rewards/margins": 0.0040840087458491325,
|
| 446 |
+
"rewards/rejected": -0.001207432011142373,
|
| 447 |
+
"step": 280
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.58,
|
| 451 |
+
"learning_rate": 4.470026884016805e-06,
|
| 452 |
+
"logits/chosen": -2.8042643070220947,
|
| 453 |
+
"logits/rejected": -2.6527392864227295,
|
| 454 |
+
"logps/chosen": -119.4009017944336,
|
| 455 |
+
"logps/rejected": -110.30342102050781,
|
| 456 |
+
"loss": 0.0015,
|
| 457 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 458 |
+
"rewards/chosen": 0.002782848197966814,
|
| 459 |
+
"rewards/margins": 0.009942631237208843,
|
| 460 |
+
"rewards/rejected": -0.007159784436225891,
|
| 461 |
+
"step": 290
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.6,
|
| 465 |
+
"learning_rate": 4.415111107797445e-06,
|
| 466 |
+
"logits/chosen": -2.963005781173706,
|
| 467 |
+
"logits/rejected": -2.809359073638916,
|
| 468 |
+
"logps/chosen": -124.54817199707031,
|
| 469 |
+
"logps/rejected": -123.4760513305664,
|
| 470 |
+
"loss": 0.0009,
|
| 471 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 472 |
+
"rewards/chosen": 0.008486522361636162,
|
| 473 |
+
"rewards/margins": 0.007400536444038153,
|
| 474 |
+
"rewards/rejected": 0.0010859851026907563,
|
| 475 |
+
"step": 300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.6,
|
| 479 |
+
"eval_logits/chosen": -2.335963249206543,
|
| 480 |
+
"eval_logits/rejected": -2.142676830291748,
|
| 481 |
+
"eval_logps/chosen": -301.29595947265625,
|
| 482 |
+
"eval_logps/rejected": -275.5566711425781,
|
| 483 |
+
"eval_loss": 0.0030309471767395735,
|
| 484 |
+
"eval_rewards/accuracies": 0.4934999942779541,
|
| 485 |
+
"eval_rewards/chosen": -0.02200666442513466,
|
| 486 |
+
"eval_rewards/margins": -0.0003654572064988315,
|
| 487 |
+
"eval_rewards/rejected": -0.021641207858920097,
|
| 488 |
+
"eval_runtime": 1423.8728,
|
| 489 |
+
"eval_samples_per_second": 1.405,
|
| 490 |
+
"eval_steps_per_second": 0.702,
|
| 491 |
+
"step": 300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.62,
|
| 495 |
+
"learning_rate": 4.357862063693486e-06,
|
| 496 |
+
"logits/chosen": -2.8513238430023193,
|
| 497 |
+
"logits/rejected": -2.672865390777588,
|
| 498 |
+
"logps/chosen": -66.75782775878906,
|
| 499 |
+
"logps/rejected": -73.05040740966797,
|
| 500 |
+
"loss": 0.001,
|
| 501 |
+
"rewards/accuracies": 0.699999988079071,
|
| 502 |
+
"rewards/chosen": -0.007263772189617157,
|
| 503 |
+
"rewards/margins": 0.007496596314013004,
|
| 504 |
+
"rewards/rejected": -0.014760365709662437,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.64,
|
| 509 |
+
"learning_rate": 4.2983495008466285e-06,
|
| 510 |
+
"logits/chosen": -2.7180657386779785,
|
| 511 |
+
"logits/rejected": -2.3951573371887207,
|
| 512 |
+
"logps/chosen": -88.17804718017578,
|
| 513 |
+
"logps/rejected": -78.4397964477539,
|
| 514 |
+
"loss": 0.0007,
|
| 515 |
+
"rewards/accuracies": 0.75,
|
| 516 |
+
"rewards/chosen": 0.009861504659056664,
|
| 517 |
+
"rewards/margins": 0.019821077585220337,
|
| 518 |
+
"rewards/rejected": -0.009959569200873375,
|
| 519 |
+
"step": 320
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.66,
|
| 523 |
+
"learning_rate": 4.236645926147493e-06,
|
| 524 |
+
"logits/chosen": -2.687565326690674,
|
| 525 |
+
"logits/rejected": -2.4989490509033203,
|
| 526 |
+
"logps/chosen": -130.00393676757812,
|
| 527 |
+
"logps/rejected": -130.13922119140625,
|
| 528 |
+
"loss": 0.0013,
|
| 529 |
+
"rewards/accuracies": 0.5,
|
| 530 |
+
"rewards/chosen": 0.04140586778521538,
|
| 531 |
+
"rewards/margins": 0.013412979431450367,
|
| 532 |
+
"rewards/rejected": 0.027992893010377884,
|
| 533 |
+
"step": 330
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.68,
|
| 537 |
+
"learning_rate": 4.172826515897146e-06,
|
| 538 |
+
"logits/chosen": -2.6931281089782715,
|
| 539 |
+
"logits/rejected": -2.3428401947021484,
|
| 540 |
+
"logps/chosen": -79.95687103271484,
|
| 541 |
+
"logps/rejected": -78.25990295410156,
|
| 542 |
+
"loss": 0.0011,
|
| 543 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 544 |
+
"rewards/chosen": 0.010445808991789818,
|
| 545 |
+
"rewards/margins": 0.013384808786213398,
|
| 546 |
+
"rewards/rejected": -0.0029389983974397182,
|
| 547 |
+
"step": 340
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.7,
|
| 551 |
+
"learning_rate": 4.106969024216348e-06,
|
| 552 |
+
"logits/chosen": -2.71099853515625,
|
| 553 |
+
"logits/rejected": -2.4201858043670654,
|
| 554 |
+
"logps/chosen": -51.87015914916992,
|
| 555 |
+
"logps/rejected": -49.29432678222656,
|
| 556 |
+
"loss": 0.0013,
|
| 557 |
+
"rewards/accuracies": 0.75,
|
| 558 |
+
"rewards/chosen": -0.0006995106232352555,
|
| 559 |
+
"rewards/margins": 0.008277077227830887,
|
| 560 |
+
"rewards/rejected": -0.008976588025689125,
|
| 561 |
+
"step": 350
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.72,
|
| 565 |
+
"learning_rate": 4.039153688314146e-06,
|
| 566 |
+
"logits/chosen": -2.80985689163208,
|
| 567 |
+
"logits/rejected": -2.5829150676727295,
|
| 568 |
+
"logps/chosen": -73.7370834350586,
|
| 569 |
+
"logps/rejected": -69.81620788574219,
|
| 570 |
+
"loss": 0.0008,
|
| 571 |
+
"rewards/accuracies": 0.75,
|
| 572 |
+
"rewards/chosen": 0.014964630827307701,
|
| 573 |
+
"rewards/margins": 0.014514106325805187,
|
| 574 |
+
"rewards/rejected": 0.0004505239485297352,
|
| 575 |
+
"step": 360
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.74,
|
| 579 |
+
"learning_rate": 3.969463130731183e-06,
|
| 580 |
+
"logits/chosen": -2.7932417392730713,
|
| 581 |
+
"logits/rejected": -2.5239012241363525,
|
| 582 |
+
"logps/chosen": -114.29008483886719,
|
| 583 |
+
"logps/rejected": -102.12156677246094,
|
| 584 |
+
"loss": 0.0011,
|
| 585 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 586 |
+
"rewards/chosen": 0.004044980742037296,
|
| 587 |
+
"rewards/margins": 0.010396437719464302,
|
| 588 |
+
"rewards/rejected": -0.006351456046104431,
|
| 589 |
+
"step": 370
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.76,
|
| 593 |
+
"learning_rate": 3.897982258676867e-06,
|
| 594 |
+
"logits/chosen": -2.7236623764038086,
|
| 595 |
+
"logits/rejected": -2.5601420402526855,
|
| 596 |
+
"logps/chosen": -84.38848876953125,
|
| 597 |
+
"logps/rejected": -85.55775451660156,
|
| 598 |
+
"loss": 0.0008,
|
| 599 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 600 |
+
"rewards/chosen": 0.012120635248720646,
|
| 601 |
+
"rewards/margins": 0.010280453599989414,
|
| 602 |
+
"rewards/rejected": 0.0018401825800538063,
|
| 603 |
+
"step": 380
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.78,
|
| 607 |
+
"learning_rate": 3.824798160583012e-06,
|
| 608 |
+
"logits/chosen": -2.8765292167663574,
|
| 609 |
+
"logits/rejected": -2.7375664710998535,
|
| 610 |
+
"logps/chosen": -107.2117919921875,
|
| 611 |
+
"logps/rejected": -103.0725326538086,
|
| 612 |
+
"loss": 0.001,
|
| 613 |
+
"rewards/accuracies": 0.550000011920929,
|
| 614 |
+
"rewards/chosen": 0.021095167845487595,
|
| 615 |
+
"rewards/margins": 0.009734300896525383,
|
| 616 |
+
"rewards/rejected": 0.011360866948962212,
|
| 617 |
+
"step": 390
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.8,
|
| 621 |
+
"learning_rate": 3.7500000000000005e-06,
|
| 622 |
+
"logits/chosen": -2.8528428077697754,
|
| 623 |
+
"logits/rejected": -2.539654493331909,
|
| 624 |
+
"logps/chosen": -86.22474670410156,
|
| 625 |
+
"logps/rejected": -74.02642822265625,
|
| 626 |
+
"loss": 0.0013,
|
| 627 |
+
"rewards/accuracies": 0.699999988079071,
|
| 628 |
+
"rewards/chosen": 0.02159132994711399,
|
| 629 |
+
"rewards/margins": 0.008513709530234337,
|
| 630 |
+
"rewards/rejected": 0.013077618554234505,
|
| 631 |
+
"step": 400
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.8,
|
| 635 |
+
"eval_logits/chosen": -2.340451955795288,
|
| 636 |
+
"eval_logits/rejected": -2.1462416648864746,
|
| 637 |
+
"eval_logps/chosen": -300.6561279296875,
|
| 638 |
+
"eval_logps/rejected": -274.8155822753906,
|
| 639 |
+
"eval_loss": 0.0034076529555022717,
|
| 640 |
+
"eval_rewards/accuracies": 0.4934999942779541,
|
| 641 |
+
"eval_rewards/chosen": -0.015608005225658417,
|
| 642 |
+
"eval_rewards/margins": -0.0013772795209661126,
|
| 643 |
+
"eval_rewards/rejected": -0.014230725355446339,
|
| 644 |
+
"eval_runtime": 1422.7868,
|
| 645 |
+
"eval_samples_per_second": 1.406,
|
| 646 |
+
"eval_steps_per_second": 0.703,
|
| 647 |
+
"step": 400
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.82,
|
| 651 |
+
"learning_rate": 3.6736789069647273e-06,
|
| 652 |
+
"logits/chosen": -2.8410146236419678,
|
| 653 |
+
"logits/rejected": -2.368011713027954,
|
| 654 |
+
"logps/chosen": -64.5174789428711,
|
| 655 |
+
"logps/rejected": -56.3655891418457,
|
| 656 |
+
"loss": 0.0015,
|
| 657 |
+
"rewards/accuracies": 0.550000011920929,
|
| 658 |
+
"rewards/chosen": 0.02277212217450142,
|
| 659 |
+
"rewards/margins": 0.015335053205490112,
|
| 660 |
+
"rewards/rejected": 0.007437069900333881,
|
| 661 |
+
"step": 410
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.84,
|
| 665 |
+
"learning_rate": 3.595927866972694e-06,
|
| 666 |
+
"logits/chosen": -2.773725986480713,
|
| 667 |
+
"logits/rejected": -2.6576743125915527,
|
| 668 |
+
"logps/chosen": -95.10247802734375,
|
| 669 |
+
"logps/rejected": -95.42603302001953,
|
| 670 |
+
"loss": 0.0007,
|
| 671 |
+
"rewards/accuracies": 0.75,
|
| 672 |
+
"rewards/chosen": 0.0311850905418396,
|
| 673 |
+
"rewards/margins": 0.010761396028101444,
|
| 674 |
+
"rewards/rejected": 0.02042369544506073,
|
| 675 |
+
"step": 420
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.86,
|
| 679 |
+
"learning_rate": 3.516841607689501e-06,
|
| 680 |
+
"logits/chosen": -2.824734926223755,
|
| 681 |
+
"logits/rejected": -2.556206226348877,
|
| 682 |
+
"logps/chosen": -108.12715911865234,
|
| 683 |
+
"logps/rejected": -96.86244201660156,
|
| 684 |
+
"loss": 0.001,
|
| 685 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 686 |
+
"rewards/chosen": 0.032226406037807465,
|
| 687 |
+
"rewards/margins": 0.0070249298587441444,
|
| 688 |
+
"rewards/rejected": 0.025201475247740746,
|
| 689 |
+
"step": 430
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.88,
|
| 693 |
+
"learning_rate": 3.436516483539781e-06,
|
| 694 |
+
"logits/chosen": -2.926262617111206,
|
| 695 |
+
"logits/rejected": -2.665631055831909,
|
| 696 |
+
"logps/chosen": -61.947792053222656,
|
| 697 |
+
"logps/rejected": -50.202911376953125,
|
| 698 |
+
"loss": 0.0006,
|
| 699 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 700 |
+
"rewards/chosen": 0.0329199843108654,
|
| 701 |
+
"rewards/margins": 0.02404285967350006,
|
| 702 |
+
"rewards/rejected": 0.00887712650001049,
|
| 703 |
+
"step": 440
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.9,
|
| 707 |
+
"learning_rate": 3.3550503583141726e-06,
|
| 708 |
+
"logits/chosen": -2.8114187717437744,
|
| 709 |
+
"logits/rejected": -2.5821549892425537,
|
| 710 |
+
"logps/chosen": -56.29111862182617,
|
| 711 |
+
"logps/rejected": -47.3681526184082,
|
| 712 |
+
"loss": 0.0013,
|
| 713 |
+
"rewards/accuracies": 0.550000011920929,
|
| 714 |
+
"rewards/chosen": 0.0194512028247118,
|
| 715 |
+
"rewards/margins": 0.006624855101108551,
|
| 716 |
+
"rewards/rejected": 0.012826347723603249,
|
| 717 |
+
"step": 450
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.92,
|
| 721 |
+
"learning_rate": 3.272542485937369e-06,
|
| 722 |
+
"logits/chosen": -2.826735019683838,
|
| 723 |
+
"logits/rejected": -2.5145363807678223,
|
| 724 |
+
"logps/chosen": -141.8760986328125,
|
| 725 |
+
"logps/rejected": -135.92344665527344,
|
| 726 |
+
"loss": 0.0016,
|
| 727 |
+
"rewards/accuracies": 0.550000011920929,
|
| 728 |
+
"rewards/chosen": 0.02360866777598858,
|
| 729 |
+
"rewards/margins": 0.004182119388133287,
|
| 730 |
+
"rewards/rejected": 0.019426550716161728,
|
| 731 |
+
"step": 460
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.94,
|
| 735 |
+
"learning_rate": 3.189093389542498e-06,
|
| 736 |
+
"logits/chosen": -2.9071967601776123,
|
| 737 |
+
"logits/rejected": -2.701315402984619,
|
| 738 |
+
"logps/chosen": -52.77317428588867,
|
| 739 |
+
"logps/rejected": -43.23090362548828,
|
| 740 |
+
"loss": 0.0008,
|
| 741 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 742 |
+
"rewards/chosen": 0.028543706983327866,
|
| 743 |
+
"rewards/margins": 0.014000201597809792,
|
| 744 |
+
"rewards/rejected": 0.014543506316840649,
|
| 745 |
+
"step": 470
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.96,
|
| 749 |
+
"learning_rate": 3.1048047389991693e-06,
|
| 750 |
+
"logits/chosen": -2.944225788116455,
|
| 751 |
+
"logits/rejected": -2.7375664710998535,
|
| 752 |
+
"logps/chosen": -123.64131927490234,
|
| 753 |
+
"logps/rejected": -121.98396301269531,
|
| 754 |
+
"loss": 0.0007,
|
| 755 |
+
"rewards/accuracies": 0.800000011920929,
|
| 756 |
+
"rewards/chosen": 0.025555819272994995,
|
| 757 |
+
"rewards/margins": 0.01265828125178814,
|
| 758 |
+
"rewards/rejected": 0.012897538021206856,
|
| 759 |
+
"step": 480
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.98,
|
| 763 |
+
"learning_rate": 3.019779227044398e-06,
|
| 764 |
+
"logits/chosen": -2.7981390953063965,
|
| 765 |
+
"logits/rejected": -2.5953376293182373,
|
| 766 |
+
"logps/chosen": -100.98303985595703,
|
| 767 |
+
"logps/rejected": -113.0623550415039,
|
| 768 |
+
"loss": 0.0009,
|
| 769 |
+
"rewards/accuracies": 0.800000011920929,
|
| 770 |
+
"rewards/chosen": 0.021701429039239883,
|
| 771 |
+
"rewards/margins": 0.013312505558133125,
|
| 772 |
+
"rewards/rejected": 0.008388923481106758,
|
| 773 |
+
"step": 490
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.0,
|
| 777 |
+
"learning_rate": 2.9341204441673267e-06,
|
| 778 |
+
"logits/chosen": -2.7481579780578613,
|
| 779 |
+
"logits/rejected": -2.5245938301086426,
|
| 780 |
+
"logps/chosen": -215.2183074951172,
|
| 781 |
+
"logps/rejected": -199.92349243164062,
|
| 782 |
+
"loss": 0.0011,
|
| 783 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 784 |
+
"rewards/chosen": 0.024366769939661026,
|
| 785 |
+
"rewards/margins": 0.014933845028281212,
|
| 786 |
+
"rewards/rejected": 0.00943292398005724,
|
| 787 |
+
"step": 500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.0,
|
| 791 |
+
"eval_logits/chosen": -2.3392491340637207,
|
| 792 |
+
"eval_logits/rejected": -2.1454310417175293,
|
| 793 |
+
"eval_logps/chosen": -304.7457275390625,
|
| 794 |
+
"eval_logps/rejected": -278.41650390625,
|
| 795 |
+
"eval_loss": 0.0036536348052322865,
|
| 796 |
+
"eval_rewards/accuracies": 0.4519999921321869,
|
| 797 |
+
"eval_rewards/chosen": -0.05650419369339943,
|
| 798 |
+
"eval_rewards/margins": -0.0062644826248288155,
|
| 799 |
+
"eval_rewards/rejected": -0.05023970827460289,
|
| 800 |
+
"eval_runtime": 1422.9482,
|
| 801 |
+
"eval_samples_per_second": 1.406,
|
| 802 |
+
"eval_steps_per_second": 0.703,
|
| 803 |
+
"step": 500
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 1.02,
|
| 807 |
+
"learning_rate": 2.847932752400164e-06,
|
| 808 |
+
"logits/chosen": -2.8486416339874268,
|
| 809 |
+
"logits/rejected": -2.635902166366577,
|
| 810 |
+
"logps/chosen": -69.38902282714844,
|
| 811 |
+
"logps/rejected": -46.49626159667969,
|
| 812 |
+
"loss": 0.0112,
|
| 813 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 814 |
+
"rewards/chosen": 0.062225647270679474,
|
| 815 |
+
"rewards/margins": 0.07529176026582718,
|
| 816 |
+
"rewards/rejected": -0.013066110201179981,
|
| 817 |
+
"step": 510
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 1.04,
|
| 821 |
+
"learning_rate": 2.761321158169134e-06,
|
| 822 |
+
"logits/chosen": -2.7715601921081543,
|
| 823 |
+
"logits/rejected": -2.531752347946167,
|
| 824 |
+
"logps/chosen": -59.361083984375,
|
| 825 |
+
"logps/rejected": -49.393150329589844,
|
| 826 |
+
"loss": 0.001,
|
| 827 |
+
"rewards/accuracies": 0.800000011920929,
|
| 828 |
+
"rewards/chosen": 0.018055101856589317,
|
| 829 |
+
"rewards/margins": 0.018152302131056786,
|
| 830 |
+
"rewards/rejected": -9.720241359900683e-05,
|
| 831 |
+
"step": 520
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.06,
|
| 835 |
+
"learning_rate": 2.6743911843603134e-06,
|
| 836 |
+
"logits/chosen": -2.88091778755188,
|
| 837 |
+
"logits/rejected": -2.714815139770508,
|
| 838 |
+
"logps/chosen": -75.8203353881836,
|
| 839 |
+
"logps/rejected": -91.77738952636719,
|
| 840 |
+
"loss": 0.0053,
|
| 841 |
+
"rewards/accuracies": 0.949999988079071,
|
| 842 |
+
"rewards/chosen": 0.03629200905561447,
|
| 843 |
+
"rewards/margins": 0.05072981119155884,
|
| 844 |
+
"rewards/rejected": -0.014437800273299217,
|
| 845 |
+
"step": 530
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.08,
|
| 849 |
+
"learning_rate": 2.587248741756253e-06,
|
| 850 |
+
"logits/chosen": -2.7633297443389893,
|
| 851 |
+
"logits/rejected": -2.4665675163269043,
|
| 852 |
+
"logps/chosen": -100.9280776977539,
|
| 853 |
+
"logps/rejected": -104.80192565917969,
|
| 854 |
+
"loss": 0.0037,
|
| 855 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 856 |
+
"rewards/chosen": 0.0258990116417408,
|
| 857 |
+
"rewards/margins": 0.028545299544930458,
|
| 858 |
+
"rewards/rejected": -0.002646292094141245,
|
| 859 |
+
"step": 540
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.1,
|
| 863 |
+
"learning_rate": 2.5e-06,
|
| 864 |
+
"logits/chosen": -2.8614132404327393,
|
| 865 |
+
"logits/rejected": -2.539499282836914,
|
| 866 |
+
"logps/chosen": -58.6580810546875,
|
| 867 |
+
"logps/rejected": -53.3350715637207,
|
| 868 |
+
"loss": 0.0207,
|
| 869 |
+
"rewards/accuracies": 0.800000011920929,
|
| 870 |
+
"rewards/chosen": 0.06129790097475052,
|
| 871 |
+
"rewards/margins": 0.07703909277915955,
|
| 872 |
+
"rewards/rejected": -0.01574121043086052,
|
| 873 |
+
"step": 550
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.12,
|
| 877 |
+
"learning_rate": 2.4127512582437486e-06,
|
| 878 |
+
"logits/chosen": -2.688095808029175,
|
| 879 |
+
"logits/rejected": -2.4665043354034424,
|
| 880 |
+
"logps/chosen": -87.18086242675781,
|
| 881 |
+
"logps/rejected": -87.13175964355469,
|
| 882 |
+
"loss": 0.0036,
|
| 883 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 884 |
+
"rewards/chosen": -0.012053154408931732,
|
| 885 |
+
"rewards/margins": 0.03777941316366196,
|
| 886 |
+
"rewards/rejected": -0.04983257129788399,
|
| 887 |
+
"step": 560
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 1.14,
|
| 891 |
+
"learning_rate": 2.325608815639687e-06,
|
| 892 |
+
"logits/chosen": -2.9195265769958496,
|
| 893 |
+
"logits/rejected": -2.6637749671936035,
|
| 894 |
+
"logps/chosen": -130.01551818847656,
|
| 895 |
+
"logps/rejected": -122.54390716552734,
|
| 896 |
+
"loss": 0.0058,
|
| 897 |
+
"rewards/accuracies": 0.699999988079071,
|
| 898 |
+
"rewards/chosen": 0.012607279233634472,
|
| 899 |
+
"rewards/margins": 0.04042843356728554,
|
| 900 |
+
"rewards/rejected": -0.02782115340232849,
|
| 901 |
+
"step": 570
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 1.16,
|
| 905 |
+
"learning_rate": 2.238678841830867e-06,
|
| 906 |
+
"logits/chosen": -2.6946258544921875,
|
| 907 |
+
"logits/rejected": -2.5128772258758545,
|
| 908 |
+
"logps/chosen": -143.2740478515625,
|
| 909 |
+
"logps/rejected": -139.3197479248047,
|
| 910 |
+
"loss": 0.0026,
|
| 911 |
+
"rewards/accuracies": 0.699999988079071,
|
| 912 |
+
"rewards/chosen": 0.0010408941889181733,
|
| 913 |
+
"rewards/margins": 0.011470312252640724,
|
| 914 |
+
"rewards/rejected": -0.010429417714476585,
|
| 915 |
+
"step": 580
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 1.18,
|
| 919 |
+
"learning_rate": 2.1520672475998374e-06,
|
| 920 |
+
"logits/chosen": -2.934413433074951,
|
| 921 |
+
"logits/rejected": -2.6382033824920654,
|
| 922 |
+
"logps/chosen": -97.65046691894531,
|
| 923 |
+
"logps/rejected": -84.37725830078125,
|
| 924 |
+
"loss": 0.0031,
|
| 925 |
+
"rewards/accuracies": 0.75,
|
| 926 |
+
"rewards/chosen": 0.02459452673792839,
|
| 927 |
+
"rewards/margins": 0.02938224747776985,
|
| 928 |
+
"rewards/rejected": -0.004787721671164036,
|
| 929 |
+
"step": 590
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.2,
|
| 933 |
+
"learning_rate": 2.0658795558326745e-06,
|
| 934 |
+
"logits/chosen": -2.9206016063690186,
|
| 935 |
+
"logits/rejected": -2.740886926651001,
|
| 936 |
+
"logps/chosen": -59.428001403808594,
|
| 937 |
+
"logps/rejected": -76.50398254394531,
|
| 938 |
+
"loss": 0.0116,
|
| 939 |
+
"rewards/accuracies": 0.800000011920929,
|
| 940 |
+
"rewards/chosen": 0.056895893067121506,
|
| 941 |
+
"rewards/margins": 0.06610508263111115,
|
| 942 |
+
"rewards/rejected": -0.00920918770134449,
|
| 943 |
+
"step": 600
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 1.2,
|
| 947 |
+
"eval_logits/chosen": -2.3448727130889893,
|
| 948 |
+
"eval_logits/rejected": -2.152658700942993,
|
| 949 |
+
"eval_logps/chosen": -301.92657470703125,
|
| 950 |
+
"eval_logps/rejected": -275.6791076660156,
|
| 951 |
+
"eval_loss": 0.004949736408889294,
|
| 952 |
+
"eval_rewards/accuracies": 0.44350001215934753,
|
| 953 |
+
"eval_rewards/chosen": -0.028312114998698235,
|
| 954 |
+
"eval_rewards/margins": -0.005446531809866428,
|
| 955 |
+
"eval_rewards/rejected": -0.02286558412015438,
|
| 956 |
+
"eval_runtime": 1423.4392,
|
| 957 |
+
"eval_samples_per_second": 1.405,
|
| 958 |
+
"eval_steps_per_second": 0.703,
|
| 959 |
+
"step": 600
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.22,
|
| 963 |
+
"learning_rate": 1.9802207729556023e-06,
|
| 964 |
+
"logits/chosen": -2.9324729442596436,
|
| 965 |
+
"logits/rejected": -2.5813498497009277,
|
| 966 |
+
"logps/chosen": -204.2996063232422,
|
| 967 |
+
"logps/rejected": -187.82083129882812,
|
| 968 |
+
"loss": 0.0123,
|
| 969 |
+
"rewards/accuracies": 0.699999988079071,
|
| 970 |
+
"rewards/chosen": -0.0035219527781009674,
|
| 971 |
+
"rewards/margins": -0.00877875555306673,
|
| 972 |
+
"rewards/rejected": 0.005256800912320614,
|
| 973 |
+
"step": 610
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.24,
|
| 977 |
+
"learning_rate": 1.895195261000831e-06,
|
| 978 |
+
"logits/chosen": -2.8246846199035645,
|
| 979 |
+
"logits/rejected": -2.683941602706909,
|
| 980 |
+
"logps/chosen": -59.7250862121582,
|
| 981 |
+
"logps/rejected": -53.9716682434082,
|
| 982 |
+
"loss": 0.0048,
|
| 983 |
+
"rewards/accuracies": 0.800000011920929,
|
| 984 |
+
"rewards/chosen": 0.055084872990846634,
|
| 985 |
+
"rewards/margins": 0.05010204762220383,
|
| 986 |
+
"rewards/rejected": 0.004982819315046072,
|
| 987 |
+
"step": 620
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1.26,
|
| 991 |
+
"learning_rate": 1.8109066104575023e-06,
|
| 992 |
+
"logits/chosen": -2.9231362342834473,
|
| 993 |
+
"logits/rejected": -2.695604085922241,
|
| 994 |
+
"logps/chosen": -56.9963264465332,
|
| 995 |
+
"logps/rejected": -63.59523391723633,
|
| 996 |
+
"loss": 0.0045,
|
| 997 |
+
"rewards/accuracies": 0.75,
|
| 998 |
+
"rewards/chosen": 0.04905426502227783,
|
| 999 |
+
"rewards/margins": 0.04801047965884209,
|
| 1000 |
+
"rewards/rejected": 0.0010437825694680214,
|
| 1001 |
+
"step": 630
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.28,
|
| 1005 |
+
"learning_rate": 1.7274575140626318e-06,
|
| 1006 |
+
"logits/chosen": -2.905579090118408,
|
| 1007 |
+
"logits/rejected": -2.642456531524658,
|
| 1008 |
+
"logps/chosen": -98.58207702636719,
|
| 1009 |
+
"logps/rejected": -93.85881805419922,
|
| 1010 |
+
"loss": 0.0013,
|
| 1011 |
+
"rewards/accuracies": 0.75,
|
| 1012 |
+
"rewards/chosen": 0.03151048347353935,
|
| 1013 |
+
"rewards/margins": 0.029562344774603844,
|
| 1014 |
+
"rewards/rejected": 0.0019481380004435778,
|
| 1015 |
+
"step": 640
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 1.3,
|
| 1019 |
+
"learning_rate": 1.6449496416858285e-06,
|
| 1020 |
+
"logits/chosen": -3.0052525997161865,
|
| 1021 |
+
"logits/rejected": -2.7315216064453125,
|
| 1022 |
+
"logps/chosen": -89.30766296386719,
|
| 1023 |
+
"logps/rejected": -104.63911437988281,
|
| 1024 |
+
"loss": 0.0338,
|
| 1025 |
+
"rewards/accuracies": 0.75,
|
| 1026 |
+
"rewards/chosen": 0.0381547287106514,
|
| 1027 |
+
"rewards/margins": 0.0724136009812355,
|
| 1028 |
+
"rewards/rejected": -0.034258872270584106,
|
| 1029 |
+
"step": 650
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.32,
|
| 1033 |
+
"learning_rate": 1.56348351646022e-06,
|
| 1034 |
+
"logits/chosen": -2.6560475826263428,
|
| 1035 |
+
"logits/rejected": -2.5114564895629883,
|
| 1036 |
+
"logps/chosen": -131.77236938476562,
|
| 1037 |
+
"logps/rejected": -127.79585266113281,
|
| 1038 |
+
"loss": 0.0026,
|
| 1039 |
+
"rewards/accuracies": 0.550000011920929,
|
| 1040 |
+
"rewards/chosen": 0.020144177600741386,
|
| 1041 |
+
"rewards/margins": 0.023866238072514534,
|
| 1042 |
+
"rewards/rejected": -0.003722056746482849,
|
| 1043 |
+
"step": 660
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.34,
|
| 1047 |
+
"learning_rate": 1.4831583923105e-06,
|
| 1048 |
+
"logits/chosen": -2.8219382762908936,
|
| 1049 |
+
"logits/rejected": -2.446211814880371,
|
| 1050 |
+
"logps/chosen": -68.78748321533203,
|
| 1051 |
+
"logps/rejected": -53.5217170715332,
|
| 1052 |
+
"loss": 0.0027,
|
| 1053 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1054 |
+
"rewards/chosen": 0.04556558281183243,
|
| 1055 |
+
"rewards/margins": 0.04576994851231575,
|
| 1056 |
+
"rewards/rejected": -0.00020437035709619522,
|
| 1057 |
+
"step": 670
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 1.36,
|
| 1061 |
+
"learning_rate": 1.4040721330273063e-06,
|
| 1062 |
+
"logits/chosen": -2.910677433013916,
|
| 1063 |
+
"logits/rejected": -2.7675867080688477,
|
| 1064 |
+
"logps/chosen": -59.946563720703125,
|
| 1065 |
+
"logps/rejected": -60.795616149902344,
|
| 1066 |
+
"loss": 0.0106,
|
| 1067 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1068 |
+
"rewards/chosen": 0.05829508230090141,
|
| 1069 |
+
"rewards/margins": 0.04981726408004761,
|
| 1070 |
+
"rewards/rejected": 0.008477816358208656,
|
| 1071 |
+
"step": 680
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.38,
|
| 1075 |
+
"learning_rate": 1.3263210930352737e-06,
|
| 1076 |
+
"logits/chosen": -2.8532004356384277,
|
| 1077 |
+
"logits/rejected": -2.6016685962677,
|
| 1078 |
+
"logps/chosen": -196.2278289794922,
|
| 1079 |
+
"logps/rejected": -204.3356170654297,
|
| 1080 |
+
"loss": 0.0076,
|
| 1081 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1082 |
+
"rewards/chosen": 0.008091023191809654,
|
| 1083 |
+
"rewards/margins": 0.023959284648299217,
|
| 1084 |
+
"rewards/rejected": -0.015868261456489563,
|
| 1085 |
+
"step": 690
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 1.4,
|
| 1089 |
+
"learning_rate": 1.2500000000000007e-06,
|
| 1090 |
+
"logits/chosen": -2.841736316680908,
|
| 1091 |
+
"logits/rejected": -2.5827274322509766,
|
| 1092 |
+
"logps/chosen": -52.04328155517578,
|
| 1093 |
+
"logps/rejected": -62.17645263671875,
|
| 1094 |
+
"loss": 0.015,
|
| 1095 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1096 |
+
"rewards/chosen": 0.026109540835022926,
|
| 1097 |
+
"rewards/margins": 0.010281763039529324,
|
| 1098 |
+
"rewards/rejected": 0.015827778726816177,
|
| 1099 |
+
"step": 700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 1.4,
|
| 1103 |
+
"eval_logits/chosen": -2.3585643768310547,
|
| 1104 |
+
"eval_logits/rejected": -2.164954662322998,
|
| 1105 |
+
"eval_logps/chosen": -301.7041320800781,
|
| 1106 |
+
"eval_logps/rejected": -275.2170104980469,
|
| 1107 |
+
"eval_loss": 0.006464843638241291,
|
| 1108 |
+
"eval_rewards/accuracies": 0.4449999928474426,
|
| 1109 |
+
"eval_rewards/chosen": -0.026087837293744087,
|
| 1110 |
+
"eval_rewards/margins": -0.007843218743801117,
|
| 1111 |
+
"eval_rewards/rejected": -0.01824462041258812,
|
| 1112 |
+
"eval_runtime": 1422.5043,
|
| 1113 |
+
"eval_samples_per_second": 1.406,
|
| 1114 |
+
"eval_steps_per_second": 0.703,
|
| 1115 |
+
"step": 700
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.42,
|
| 1119 |
+
"learning_rate": 1.1752018394169882e-06,
|
| 1120 |
+
"logits/chosen": -2.8012290000915527,
|
| 1121 |
+
"logits/rejected": -2.348876476287842,
|
| 1122 |
+
"logps/chosen": -50.55641555786133,
|
| 1123 |
+
"logps/rejected": -46.36492156982422,
|
| 1124 |
+
"loss": 0.0033,
|
| 1125 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1126 |
+
"rewards/chosen": 0.0446331761777401,
|
| 1127 |
+
"rewards/margins": 0.04935276508331299,
|
| 1128 |
+
"rewards/rejected": -0.004719586111605167,
|
| 1129 |
+
"step": 710
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.44,
|
| 1133 |
+
"learning_rate": 1.1020177413231334e-06,
|
| 1134 |
+
"logits/chosen": -2.766397476196289,
|
| 1135 |
+
"logits/rejected": -2.423826217651367,
|
| 1136 |
+
"logps/chosen": -111.70279693603516,
|
| 1137 |
+
"logps/rejected": -129.45628356933594,
|
| 1138 |
+
"loss": 0.0049,
|
| 1139 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1140 |
+
"rewards/chosen": 0.04479371756315231,
|
| 1141 |
+
"rewards/margins": 0.06533690541982651,
|
| 1142 |
+
"rewards/rejected": -0.020543191581964493,
|
| 1143 |
+
"step": 720
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.46,
|
| 1147 |
+
"learning_rate": 1.0305368692688175e-06,
|
| 1148 |
+
"logits/chosen": -2.9610862731933594,
|
| 1149 |
+
"logits/rejected": -2.7894320487976074,
|
| 1150 |
+
"logps/chosen": -86.23368835449219,
|
| 1151 |
+
"logps/rejected": -86.16008758544922,
|
| 1152 |
+
"loss": 0.0027,
|
| 1153 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1154 |
+
"rewards/chosen": 0.04561948776245117,
|
| 1155 |
+
"rewards/margins": 0.04706469178199768,
|
| 1156 |
+
"rewards/rejected": -0.0014452062314376235,
|
| 1157 |
+
"step": 730
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.48,
|
| 1161 |
+
"learning_rate": 9.608463116858544e-07,
|
| 1162 |
+
"logits/chosen": -2.8900036811828613,
|
| 1163 |
+
"logits/rejected": -2.4443981647491455,
|
| 1164 |
+
"logps/chosen": -70.54168701171875,
|
| 1165 |
+
"logps/rejected": -50.89817428588867,
|
| 1166 |
+
"loss": 0.0013,
|
| 1167 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1168 |
+
"rewards/chosen": 0.031780801713466644,
|
| 1169 |
+
"rewards/margins": 0.038311202079057693,
|
| 1170 |
+
"rewards/rejected": -0.006530401296913624,
|
| 1171 |
+
"step": 740
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.5,
|
| 1175 |
+
"learning_rate": 8.930309757836517e-07,
|
| 1176 |
+
"logits/chosen": -2.9060871601104736,
|
| 1177 |
+
"logits/rejected": -2.524108648300171,
|
| 1178 |
+
"logps/chosen": -61.922813415527344,
|
| 1179 |
+
"logps/rejected": -65.15422821044922,
|
| 1180 |
+
"loss": 0.0056,
|
| 1181 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1182 |
+
"rewards/chosen": 0.0495627298951149,
|
| 1183 |
+
"rewards/margins": 0.04960542917251587,
|
| 1184 |
+
"rewards/rejected": -4.269964847480878e-05,
|
| 1185 |
+
"step": 750
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 1.52,
|
| 1189 |
+
"learning_rate": 8.271734841028553e-07,
|
| 1190 |
+
"logits/chosen": -2.644831657409668,
|
| 1191 |
+
"logits/rejected": -2.305298328399658,
|
| 1192 |
+
"logps/chosen": -67.29566955566406,
|
| 1193 |
+
"logps/rejected": -63.685523986816406,
|
| 1194 |
+
"loss": 0.0035,
|
| 1195 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1196 |
+
"rewards/chosen": 0.04994013160467148,
|
| 1197 |
+
"rewards/margins": 0.052011966705322266,
|
| 1198 |
+
"rewards/rejected": -0.002071837428957224,
|
| 1199 |
+
"step": 760
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.54,
|
| 1203 |
+
"learning_rate": 7.633540738525066e-07,
|
| 1204 |
+
"logits/chosen": -2.9390599727630615,
|
| 1205 |
+
"logits/rejected": -2.688645839691162,
|
| 1206 |
+
"logps/chosen": -65.12007904052734,
|
| 1207 |
+
"logps/rejected": -62.56877517700195,
|
| 1208 |
+
"loss": 0.1119,
|
| 1209 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1210 |
+
"rewards/chosen": 0.05300932377576828,
|
| 1211 |
+
"rewards/margins": 0.12731197476387024,
|
| 1212 |
+
"rewards/rejected": -0.07430265843868256,
|
| 1213 |
+
"step": 770
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.56,
|
| 1217 |
+
"learning_rate": 7.016504991533727e-07,
|
| 1218 |
+
"logits/chosen": -2.625112295150757,
|
| 1219 |
+
"logits/rejected": -2.447942018508911,
|
| 1220 |
+
"logps/chosen": -58.0603141784668,
|
| 1221 |
+
"logps/rejected": -50.18201446533203,
|
| 1222 |
+
"loss": 0.0028,
|
| 1223 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1224 |
+
"rewards/chosen": 0.009573986753821373,
|
| 1225 |
+
"rewards/margins": 0.0028710602782666683,
|
| 1226 |
+
"rewards/rejected": 0.006702926941215992,
|
| 1227 |
+
"step": 780
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 1.58,
|
| 1231 |
+
"learning_rate": 6.421379363065142e-07,
|
| 1232 |
+
"logits/chosen": -2.982449769973755,
|
| 1233 |
+
"logits/rejected": -2.7615859508514404,
|
| 1234 |
+
"logps/chosen": -51.4494743347168,
|
| 1235 |
+
"logps/rejected": -57.383644104003906,
|
| 1236 |
+
"loss": 0.0061,
|
| 1237 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1238 |
+
"rewards/chosen": 0.05977553874254227,
|
| 1239 |
+
"rewards/margins": 0.05702256038784981,
|
| 1240 |
+
"rewards/rejected": 0.002752984408289194,
|
| 1241 |
+
"step": 790
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.6,
|
| 1245 |
+
"learning_rate": 5.848888922025553e-07,
|
| 1246 |
+
"logits/chosen": -2.7822391986846924,
|
| 1247 |
+
"logits/rejected": -2.5350170135498047,
|
| 1248 |
+
"logps/chosen": -43.52353286743164,
|
| 1249 |
+
"logps/rejected": -54.28175735473633,
|
| 1250 |
+
"loss": 0.0009,
|
| 1251 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1252 |
+
"rewards/chosen": 0.03572973236441612,
|
| 1253 |
+
"rewards/margins": 0.035435039550065994,
|
| 1254 |
+
"rewards/rejected": 0.0002946966851595789,
|
| 1255 |
+
"step": 800
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.6,
|
| 1259 |
+
"eval_logits/chosen": -2.3689210414886475,
|
| 1260 |
+
"eval_logits/rejected": -2.174628496170044,
|
| 1261 |
+
"eval_logps/chosen": -298.3010559082031,
|
| 1262 |
+
"eval_logps/rejected": -272.15399169921875,
|
| 1263 |
+
"eval_loss": 0.006914817728102207,
|
| 1264 |
+
"eval_rewards/accuracies": 0.47200000286102295,
|
| 1265 |
+
"eval_rewards/chosen": 0.007942860946059227,
|
| 1266 |
+
"eval_rewards/margins": -0.0044423178769648075,
|
| 1267 |
+
"eval_rewards/rejected": 0.012385179288685322,
|
| 1268 |
+
"eval_runtime": 1422.6658,
|
| 1269 |
+
"eval_samples_per_second": 1.406,
|
| 1270 |
+
"eval_steps_per_second": 0.703,
|
| 1271 |
+
"step": 800
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 1.62,
|
| 1275 |
+
"learning_rate": 5.299731159831953e-07,
|
| 1276 |
+
"logits/chosen": -2.868394136428833,
|
| 1277 |
+
"logits/rejected": -2.6725270748138428,
|
| 1278 |
+
"logps/chosen": -113.2739486694336,
|
| 1279 |
+
"logps/rejected": -118.2945327758789,
|
| 1280 |
+
"loss": 0.0086,
|
| 1281 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1282 |
+
"rewards/chosen": 0.050359614193439484,
|
| 1283 |
+
"rewards/margins": 0.05471315234899521,
|
| 1284 |
+
"rewards/rejected": -0.004353542346507311,
|
| 1285 |
+
"step": 810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 1.64,
|
| 1289 |
+
"learning_rate": 4.774575140626317e-07,
|
| 1290 |
+
"logits/chosen": -2.7905325889587402,
|
| 1291 |
+
"logits/rejected": -2.5390892028808594,
|
| 1292 |
+
"logps/chosen": -61.987098693847656,
|
| 1293 |
+
"logps/rejected": -72.47862243652344,
|
| 1294 |
+
"loss": 0.0051,
|
| 1295 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1296 |
+
"rewards/chosen": 0.035242609679698944,
|
| 1297 |
+
"rewards/margins": 0.04513431712985039,
|
| 1298 |
+
"rewards/rejected": -0.009891709312796593,
|
| 1299 |
+
"step": 820
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.66,
|
| 1303 |
+
"learning_rate": 4.27406068612396e-07,
|
| 1304 |
+
"logits/chosen": -2.744027614593506,
|
| 1305 |
+
"logits/rejected": -2.451847553253174,
|
| 1306 |
+
"logps/chosen": -94.64505767822266,
|
| 1307 |
+
"logps/rejected": -86.02244567871094,
|
| 1308 |
+
"loss": 0.0009,
|
| 1309 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1310 |
+
"rewards/chosen": 0.022839754819869995,
|
| 1311 |
+
"rewards/margins": 0.03421180695295334,
|
| 1312 |
+
"rewards/rejected": -0.011372053064405918,
|
| 1313 |
+
"step": 830
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 1.68,
|
| 1317 |
+
"learning_rate": 3.798797596089351e-07,
|
| 1318 |
+
"logits/chosen": -2.88804292678833,
|
| 1319 |
+
"logits/rejected": -2.348177194595337,
|
| 1320 |
+
"logps/chosen": -62.28667068481445,
|
| 1321 |
+
"logps/rejected": -47.92333221435547,
|
| 1322 |
+
"loss": 0.0024,
|
| 1323 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1324 |
+
"rewards/chosen": 0.031227191910147667,
|
| 1325 |
+
"rewards/margins": 0.039407771080732346,
|
| 1326 |
+
"rewards/rejected": -0.008180581033229828,
|
| 1327 |
+
"step": 840
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 1.7,
|
| 1331 |
+
"learning_rate": 3.3493649053890325e-07,
|
| 1332 |
+
"logits/chosen": -2.8707077503204346,
|
| 1333 |
+
"logits/rejected": -2.6572961807250977,
|
| 1334 |
+
"logps/chosen": -97.13041687011719,
|
| 1335 |
+
"logps/rejected": -102.47623443603516,
|
| 1336 |
+
"loss": 0.0026,
|
| 1337 |
+
"rewards/accuracies": 0.75,
|
| 1338 |
+
"rewards/chosen": 0.015258235856890678,
|
| 1339 |
+
"rewards/margins": 0.032782141119241714,
|
| 1340 |
+
"rewards/rejected": -0.017523903399705887,
|
| 1341 |
+
"step": 850
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.72,
|
| 1345 |
+
"learning_rate": 2.9263101785268253e-07,
|
| 1346 |
+
"logits/chosen": -2.6006197929382324,
|
| 1347 |
+
"logits/rejected": -2.57403826713562,
|
| 1348 |
+
"logps/chosen": -182.5235595703125,
|
| 1349 |
+
"logps/rejected": -166.66905212402344,
|
| 1350 |
+
"loss": 0.001,
|
| 1351 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1352 |
+
"rewards/chosen": 0.0023535334039479494,
|
| 1353 |
+
"rewards/margins": 0.01678496040403843,
|
| 1354 |
+
"rewards/rejected": -0.0144314244389534,
|
| 1355 |
+
"step": 860
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 1.74,
|
| 1359 |
+
"learning_rate": 2.53014884252083e-07,
|
| 1360 |
+
"logits/chosen": -2.9139671325683594,
|
| 1361 |
+
"logits/rejected": -2.8268768787384033,
|
| 1362 |
+
"logps/chosen": -132.52719116210938,
|
| 1363 |
+
"logps/rejected": -130.03152465820312,
|
| 1364 |
+
"loss": 0.0014,
|
| 1365 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1366 |
+
"rewards/chosen": 0.038164813071489334,
|
| 1367 |
+
"rewards/margins": 0.03188065439462662,
|
| 1368 |
+
"rewards/rejected": 0.006284159608185291,
|
| 1369 |
+
"step": 870
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 1.76,
|
| 1373 |
+
"learning_rate": 2.1613635589349756e-07,
|
| 1374 |
+
"logits/chosen": -2.772710084915161,
|
| 1375 |
+
"logits/rejected": -2.3980655670166016,
|
| 1376 |
+
"logps/chosen": -47.05355453491211,
|
| 1377 |
+
"logps/rejected": -44.157875061035156,
|
| 1378 |
+
"loss": 0.0034,
|
| 1379 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1380 |
+
"rewards/chosen": 0.03233030438423157,
|
| 1381 |
+
"rewards/margins": 0.03725982457399368,
|
| 1382 |
+
"rewards/rejected": -0.00492951599881053,
|
| 1383 |
+
"step": 880
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.78,
|
| 1387 |
+
"learning_rate": 1.8204036358303173e-07,
|
| 1388 |
+
"logits/chosen": -2.9192633628845215,
|
| 1389 |
+
"logits/rejected": -2.627864360809326,
|
| 1390 |
+
"logps/chosen": -70.42695617675781,
|
| 1391 |
+
"logps/rejected": -99.09819030761719,
|
| 1392 |
+
"loss": 0.1069,
|
| 1393 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1394 |
+
"rewards/chosen": 0.06590764224529266,
|
| 1395 |
+
"rewards/margins": 0.1522078961133957,
|
| 1396 |
+
"rewards/rejected": -0.08630025386810303,
|
| 1397 |
+
"step": 890
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 1.8,
|
| 1401 |
+
"learning_rate": 1.507684480352292e-07,
|
| 1402 |
+
"logits/chosen": -2.7338740825653076,
|
| 1403 |
+
"logits/rejected": -2.4327640533447266,
|
| 1404 |
+
"logps/chosen": -48.297508239746094,
|
| 1405 |
+
"logps/rejected": -40.352996826171875,
|
| 1406 |
+
"loss": 0.0109,
|
| 1407 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1408 |
+
"rewards/chosen": 0.037285320460796356,
|
| 1409 |
+
"rewards/margins": 0.0314871221780777,
|
| 1410 |
+
"rewards/rejected": 0.005798191763460636,
|
| 1411 |
+
"step": 900
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 1.8,
|
| 1415 |
+
"eval_logits/chosen": -2.368154764175415,
|
| 1416 |
+
"eval_logits/rejected": -2.1738698482513428,
|
| 1417 |
+
"eval_logps/chosen": -298.8583068847656,
|
| 1418 |
+
"eval_logps/rejected": -272.5880432128906,
|
| 1419 |
+
"eval_loss": 0.006868657190352678,
|
| 1420 |
+
"eval_rewards/accuracies": 0.4569999873638153,
|
| 1421 |
+
"eval_rewards/chosen": 0.0023701086174696684,
|
| 1422 |
+
"eval_rewards/margins": -0.005675030872225761,
|
| 1423 |
+
"eval_rewards/rejected": 0.008045138791203499,
|
| 1424 |
+
"eval_runtime": 1423.7285,
|
| 1425 |
+
"eval_samples_per_second": 1.405,
|
| 1426 |
+
"eval_steps_per_second": 0.702,
|
| 1427 |
+
"step": 900
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 1.82,
|
| 1431 |
+
"learning_rate": 1.223587092621162e-07,
|
| 1432 |
+
"logits/chosen": -3.062129497528076,
|
| 1433 |
+
"logits/rejected": -2.726337432861328,
|
| 1434 |
+
"logps/chosen": -103.9426040649414,
|
| 1435 |
+
"logps/rejected": -109.22408294677734,
|
| 1436 |
+
"loss": 0.0132,
|
| 1437 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1438 |
+
"rewards/chosen": 0.01760241389274597,
|
| 1439 |
+
"rewards/margins": 0.04251215234398842,
|
| 1440 |
+
"rewards/rejected": -0.024909736588597298,
|
| 1441 |
+
"step": 910
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 1.84,
|
| 1445 |
+
"learning_rate": 9.684576015420277e-08,
|
| 1446 |
+
"logits/chosen": -2.8683624267578125,
|
| 1447 |
+
"logits/rejected": -2.5635323524475098,
|
| 1448 |
+
"logps/chosen": -132.88455200195312,
|
| 1449 |
+
"logps/rejected": -131.37045288085938,
|
| 1450 |
+
"loss": 0.0063,
|
| 1451 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1452 |
+
"rewards/chosen": 0.03422384709119797,
|
| 1453 |
+
"rewards/margins": 0.05052335187792778,
|
| 1454 |
+
"rewards/rejected": -0.016299499198794365,
|
| 1455 |
+
"step": 920
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.86,
|
| 1459 |
+
"learning_rate": 7.426068431000883e-08,
|
| 1460 |
+
"logits/chosen": -2.763186454772949,
|
| 1461 |
+
"logits/rejected": -2.4577767848968506,
|
| 1462 |
+
"logps/chosen": -62.349082946777344,
|
| 1463 |
+
"logps/rejected": -48.44455337524414,
|
| 1464 |
+
"loss": 0.0027,
|
| 1465 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1466 |
+
"rewards/chosen": 0.03767557442188263,
|
| 1467 |
+
"rewards/margins": 0.04414733499288559,
|
| 1468 |
+
"rewards/rejected": -0.006471761967986822,
|
| 1469 |
+
"step": 930
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.88,
|
| 1473 |
+
"learning_rate": 5.463099816548578e-08,
|
| 1474 |
+
"logits/chosen": -2.749911308288574,
|
| 1475 |
+
"logits/rejected": -2.5365560054779053,
|
| 1476 |
+
"logps/chosen": -109.21867370605469,
|
| 1477 |
+
"logps/rejected": -97.2970962524414,
|
| 1478 |
+
"loss": 0.0013,
|
| 1479 |
+
"rewards/accuracies": 0.75,
|
| 1480 |
+
"rewards/chosen": 0.03286752849817276,
|
| 1481 |
+
"rewards/margins": 0.033335305750370026,
|
| 1482 |
+
"rewards/rejected": -0.00046777556417509913,
|
| 1483 |
+
"step": 940
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.9,
|
| 1487 |
+
"learning_rate": 3.798061746947995e-08,
|
| 1488 |
+
"logits/chosen": -2.9467415809631348,
|
| 1489 |
+
"logits/rejected": -2.6370301246643066,
|
| 1490 |
+
"logps/chosen": -142.62974548339844,
|
| 1491 |
+
"logps/rejected": -140.1845703125,
|
| 1492 |
+
"loss": 0.0015,
|
| 1493 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1494 |
+
"rewards/chosen": 0.03325565904378891,
|
| 1495 |
+
"rewards/margins": 0.027630716562271118,
|
| 1496 |
+
"rewards/rejected": 0.005624941550195217,
|
| 1497 |
+
"step": 950
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 1.92,
|
| 1501 |
+
"learning_rate": 2.4329828146074096e-08,
|
| 1502 |
+
"logits/chosen": -2.7907421588897705,
|
| 1503 |
+
"logits/rejected": -2.5240225791931152,
|
| 1504 |
+
"logps/chosen": -112.5775146484375,
|
| 1505 |
+
"logps/rejected": -115.86857604980469,
|
| 1506 |
+
"loss": 0.0156,
|
| 1507 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1508 |
+
"rewards/chosen": 0.012037028558552265,
|
| 1509 |
+
"rewards/margins": -0.006716550327837467,
|
| 1510 |
+
"rewards/rejected": 0.01875358074903488,
|
| 1511 |
+
"step": 960
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 1.94,
|
| 1515 |
+
"learning_rate": 1.3695261579316776e-08,
|
| 1516 |
+
"logits/chosen": -3.0265414714813232,
|
| 1517 |
+
"logits/rejected": -2.6483359336853027,
|
| 1518 |
+
"logps/chosen": -85.69535064697266,
|
| 1519 |
+
"logps/rejected": -82.9826889038086,
|
| 1520 |
+
"loss": 0.0296,
|
| 1521 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1522 |
+
"rewards/chosen": 0.013691084459424019,
|
| 1523 |
+
"rewards/margins": 0.002223056508228183,
|
| 1524 |
+
"rewards/rejected": 0.011468032374978065,
|
| 1525 |
+
"step": 970
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 1.96,
|
| 1529 |
+
"learning_rate": 6.089874350439507e-09,
|
| 1530 |
+
"logits/chosen": -2.6236181259155273,
|
| 1531 |
+
"logits/rejected": -2.498073101043701,
|
| 1532 |
+
"logps/chosen": -81.69920349121094,
|
| 1533 |
+
"logps/rejected": -90.76789855957031,
|
| 1534 |
+
"loss": 0.0062,
|
| 1535 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1536 |
+
"rewards/chosen": 0.018706561997532845,
|
| 1537 |
+
"rewards/margins": 0.04268919676542282,
|
| 1538 |
+
"rewards/rejected": -0.023982631042599678,
|
| 1539 |
+
"step": 980
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 1.98,
|
| 1543 |
+
"learning_rate": 1.5229324522605949e-09,
|
| 1544 |
+
"logits/chosen": -2.952378034591675,
|
| 1545 |
+
"logits/rejected": -2.7012767791748047,
|
| 1546 |
+
"logps/chosen": -114.75297546386719,
|
| 1547 |
+
"logps/rejected": -98.20231628417969,
|
| 1548 |
+
"loss": 0.0073,
|
| 1549 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1550 |
+
"rewards/chosen": 0.020036840811371803,
|
| 1551 |
+
"rewards/margins": 0.031693797558546066,
|
| 1552 |
+
"rewards/rejected": -0.011656956747174263,
|
| 1553 |
+
"step": 990
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 2.0,
|
| 1557 |
+
"learning_rate": 0.0,
|
| 1558 |
+
"logits/chosen": -2.9289968013763428,
|
| 1559 |
+
"logits/rejected": -2.6982641220092773,
|
| 1560 |
+
"logps/chosen": -100.9759750366211,
|
| 1561 |
+
"logps/rejected": -94.3797607421875,
|
| 1562 |
+
"loss": 0.0015,
|
| 1563 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1564 |
+
"rewards/chosen": 0.02024517022073269,
|
| 1565 |
+
"rewards/margins": 0.02688099816441536,
|
| 1566 |
+
"rewards/rejected": -0.006635826081037521,
|
| 1567 |
+
"step": 1000
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 2.0,
|
| 1571 |
+
"eval_logits/chosen": -2.3691599369049072,
|
| 1572 |
+
"eval_logits/rejected": -2.1749157905578613,
|
| 1573 |
+
"eval_logps/chosen": -298.8498229980469,
|
| 1574 |
+
"eval_logps/rejected": -272.58660888671875,
|
| 1575 |
+
"eval_loss": 0.006900906562805176,
|
| 1576 |
+
"eval_rewards/accuracies": 0.4595000147819519,
|
| 1577 |
+
"eval_rewards/chosen": 0.002455136040225625,
|
| 1578 |
+
"eval_rewards/margins": -0.005604185629636049,
|
| 1579 |
+
"eval_rewards/rejected": 0.008059320971369743,
|
| 1580 |
+
"eval_runtime": 1422.3751,
|
| 1581 |
+
"eval_samples_per_second": 1.406,
|
| 1582 |
+
"eval_steps_per_second": 0.703,
|
| 1583 |
+
"step": 1000
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 2.0,
|
| 1587 |
+
"step": 1000,
|
| 1588 |
+
"total_flos": 0.0,
|
| 1589 |
+
"train_loss": 0.005910432943725027,
|
| 1590 |
+
"train_runtime": 19053.6119,
|
| 1591 |
+
"train_samples_per_second": 0.105,
|
| 1592 |
+
"train_steps_per_second": 0.052
|
| 1593 |
+
}
|
| 1594 |
+
],
|
| 1595 |
+
"logging_steps": 10,
|
| 1596 |
+
"max_steps": 1000,
|
| 1597 |
+
"num_input_tokens_seen": 0,
|
| 1598 |
+
"num_train_epochs": 2,
|
| 1599 |
+
"save_steps": 100,
|
| 1600 |
+
"total_flos": 0.0,
|
| 1601 |
+
"train_batch_size": 1,
|
| 1602 |
+
"trial_name": null,
|
| 1603 |
+
"trial_params": null
|
| 1604 |
+
}
|