Model save
Browse files- README.md +83 -0
- adapter_model.safetensors +1 -1
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1626 -0
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
tags:
|
| 4 |
+
- trl
|
| 5 |
+
- dpo
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
base_model: data/zephyr-7b-sft-qlora-merged
|
| 8 |
+
model-index:
|
| 9 |
+
- name: zephyr-7b-dpo-qlora
|
| 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 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/statking/huggingface/runs/qxp2vmm7)
|
| 17 |
+
# zephyr-7b-dpo-qlora
|
| 18 |
+
|
| 19 |
+
This model was trained from scratch on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.4933
|
| 22 |
+
- Rewards/chosen: -2.1713
|
| 23 |
+
- Rewards/rejected: -3.1801
|
| 24 |
+
- Rewards/accuracies: 0.7738
|
| 25 |
+
- Rewards/margins: 1.0088
|
| 26 |
+
- Logps/rejected: -564.8470
|
| 27 |
+
- Logps/chosen: -483.4024
|
| 28 |
+
- Logits/rejected: -1.4105
|
| 29 |
+
- Logits/chosen: -1.4778
|
| 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: 4
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 4
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 32
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 1
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.6185 | 0.1047 | 100 | 0.6240 | -0.3010 | -0.5396 | 0.6964 | 0.2387 | -300.7997 | -296.3736 | -2.2954 | -2.3537 |
|
| 67 |
+
| 0.5724 | 0.2094 | 200 | 0.5692 | -0.8434 | -1.3284 | 0.7302 | 0.4850 | -379.6750 | -350.6113 | -2.2448 | -2.2930 |
|
| 68 |
+
| 0.5366 | 0.3141 | 300 | 0.5249 | -1.6887 | -2.4863 | 0.7639 | 0.7976 | -495.4648 | -435.1429 | -1.6220 | -1.6850 |
|
| 69 |
+
| 0.5397 | 0.4187 | 400 | 0.5253 | -1.2998 | -1.9923 | 0.7698 | 0.6925 | -446.0619 | -396.2537 | -1.7586 | -1.8144 |
|
| 70 |
+
| 0.5003 | 0.5234 | 500 | 0.5013 | -1.9982 | -2.9207 | 0.7659 | 0.9226 | -538.9065 | -466.0909 | -1.6049 | -1.6682 |
|
| 71 |
+
| 0.4835 | 0.6281 | 600 | 0.5027 | -2.5699 | -3.5168 | 0.7560 | 0.9470 | -598.5182 | -523.2593 | -1.3417 | -1.4125 |
|
| 72 |
+
| 0.4715 | 0.7328 | 700 | 0.4956 | -2.1902 | -3.1936 | 0.7679 | 1.0035 | -566.1955 | -485.2894 | -1.3782 | -1.4480 |
|
| 73 |
+
| 0.4898 | 0.8375 | 800 | 0.4948 | -2.0401 | -3.0116 | 0.7698 | 0.9715 | -547.9974 | -470.2821 | -1.4275 | -1.4946 |
|
| 74 |
+
| 0.4785 | 0.9422 | 900 | 0.4933 | -2.1713 | -3.1801 | 0.7738 | 1.0088 | -564.8470 | -483.4024 | -1.4105 | -1.4778 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- PEFT 0.10.0
|
| 80 |
+
- Transformers 4.41.0.dev0
|
| 81 |
+
- Pytorch 2.3.0+cu121
|
| 82 |
+
- Datasets 2.19.1
|
| 83 |
+
- Tokenizers 0.19.1
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671150064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0959395381e8b38e8e47659e117c11d160900e65148219841bc9a27c24ed8c6a
|
| 3 |
size 671150064
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9997382884061764,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5333428270529702,
|
| 5 |
+
"train_runtime": 30179.1373,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 2.026,
|
| 8 |
+
"train_steps_per_second": 0.032
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9997382884061764,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.5333428270529702,
|
| 5 |
+
"train_runtime": 30179.1373,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 2.026,
|
| 8 |
+
"train_steps_per_second": 0.032
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1626 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9997382884061764,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 955,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0010468463752944255,
|
| 13 |
+
"grad_norm": 1.1290596695646078,
|
| 14 |
+
"learning_rate": 5.208333333333333e-08,
|
| 15 |
+
"logits/chosen": -2.659804344177246,
|
| 16 |
+
"logits/rejected": -2.5501840114593506,
|
| 17 |
+
"logps/chosen": -300.040771484375,
|
| 18 |
+
"logps/rejected": -255.0087127685547,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.010468463752944255,
|
| 28 |
+
"grad_norm": 1.014414920556849,
|
| 29 |
+
"learning_rate": 5.208333333333334e-07,
|
| 30 |
+
"logits/chosen": -2.438154935836792,
|
| 31 |
+
"logits/rejected": -2.421139717102051,
|
| 32 |
+
"logps/chosen": -277.2698669433594,
|
| 33 |
+
"logps/rejected": -256.98931884765625,
|
| 34 |
+
"loss": 0.6932,
|
| 35 |
+
"rewards/accuracies": 0.4930555522441864,
|
| 36 |
+
"rewards/chosen": 8.686767250765115e-05,
|
| 37 |
+
"rewards/margins": 0.0004906932590529323,
|
| 38 |
+
"rewards/rejected": -0.0004038256302010268,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02093692750588851,
|
| 43 |
+
"grad_norm": 1.1593543431279631,
|
| 44 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 45 |
+
"logits/chosen": -2.4331612586975098,
|
| 46 |
+
"logits/rejected": -2.344513177871704,
|
| 47 |
+
"logps/chosen": -281.2090759277344,
|
| 48 |
+
"logps/rejected": -260.51007080078125,
|
| 49 |
+
"loss": 0.6926,
|
| 50 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 51 |
+
"rewards/chosen": 0.0008262035553343594,
|
| 52 |
+
"rewards/margins": 0.0013498691841959953,
|
| 53 |
+
"rewards/rejected": -0.0005236656288616359,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.031405391258832765,
|
| 58 |
+
"grad_norm": 1.1658712617217375,
|
| 59 |
+
"learning_rate": 1.5625e-06,
|
| 60 |
+
"logits/chosen": -2.492827892303467,
|
| 61 |
+
"logits/rejected": -2.421010971069336,
|
| 62 |
+
"logps/chosen": -286.5789489746094,
|
| 63 |
+
"logps/rejected": -267.0865783691406,
|
| 64 |
+
"loss": 0.6909,
|
| 65 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 66 |
+
"rewards/chosen": 0.0038475811015814543,
|
| 67 |
+
"rewards/margins": 0.004234342835843563,
|
| 68 |
+
"rewards/rejected": -0.0003867618797812611,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04187385501177702,
|
| 73 |
+
"grad_norm": 1.1449870340989134,
|
| 74 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 75 |
+
"logits/chosen": -2.486035108566284,
|
| 76 |
+
"logits/rejected": -2.4186811447143555,
|
| 77 |
+
"logps/chosen": -274.2925720214844,
|
| 78 |
+
"logps/rejected": -260.7481384277344,
|
| 79 |
+
"loss": 0.688,
|
| 80 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 81 |
+
"rewards/chosen": 0.011111991479992867,
|
| 82 |
+
"rewards/margins": 0.01205919124186039,
|
| 83 |
+
"rewards/rejected": -0.0009472008096054196,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.05234231876472128,
|
| 88 |
+
"grad_norm": 1.1126402014071601,
|
| 89 |
+
"learning_rate": 2.604166666666667e-06,
|
| 90 |
+
"logits/chosen": -2.4758384227752686,
|
| 91 |
+
"logits/rejected": -2.404066562652588,
|
| 92 |
+
"logps/chosen": -233.27294921875,
|
| 93 |
+
"logps/rejected": -210.6988525390625,
|
| 94 |
+
"loss": 0.6835,
|
| 95 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 96 |
+
"rewards/chosen": 0.023658782243728638,
|
| 97 |
+
"rewards/margins": 0.025034388527274132,
|
| 98 |
+
"rewards/rejected": -0.0013756046537309885,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.06281078251766553,
|
| 103 |
+
"grad_norm": 1.2225282330679852,
|
| 104 |
+
"learning_rate": 3.125e-06,
|
| 105 |
+
"logits/chosen": -2.4660396575927734,
|
| 106 |
+
"logits/rejected": -2.3986904621124268,
|
| 107 |
+
"logps/chosen": -268.0579528808594,
|
| 108 |
+
"logps/rejected": -230.85205078125,
|
| 109 |
+
"loss": 0.6753,
|
| 110 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 111 |
+
"rewards/chosen": 0.029377218335866928,
|
| 112 |
+
"rewards/margins": 0.03872579708695412,
|
| 113 |
+
"rewards/rejected": -0.009348581545054913,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.07327924627060979,
|
| 118 |
+
"grad_norm": 1.3167193147118952,
|
| 119 |
+
"learning_rate": 3.6458333333333333e-06,
|
| 120 |
+
"logits/chosen": -2.3562026023864746,
|
| 121 |
+
"logits/rejected": -2.327336072921753,
|
| 122 |
+
"logps/chosen": -256.792724609375,
|
| 123 |
+
"logps/rejected": -262.48309326171875,
|
| 124 |
+
"loss": 0.6596,
|
| 125 |
+
"rewards/accuracies": 0.78125,
|
| 126 |
+
"rewards/chosen": 0.008793818764388561,
|
| 127 |
+
"rewards/margins": 0.08649717271327972,
|
| 128 |
+
"rewards/rejected": -0.07770337164402008,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.08374771002355404,
|
| 133 |
+
"grad_norm": 1.8338431766961225,
|
| 134 |
+
"learning_rate": 4.166666666666667e-06,
|
| 135 |
+
"logits/chosen": -2.4636688232421875,
|
| 136 |
+
"logits/rejected": -2.3472182750701904,
|
| 137 |
+
"logps/chosen": -264.05975341796875,
|
| 138 |
+
"logps/rejected": -255.5188446044922,
|
| 139 |
+
"loss": 0.6425,
|
| 140 |
+
"rewards/accuracies": 0.71875,
|
| 141 |
+
"rewards/chosen": -0.10224815458059311,
|
| 142 |
+
"rewards/margins": 0.11396624147891998,
|
| 143 |
+
"rewards/rejected": -0.2162143886089325,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.0942161737764983,
|
| 148 |
+
"grad_norm": 1.8214883674875806,
|
| 149 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 150 |
+
"logits/chosen": -2.3876922130584717,
|
| 151 |
+
"logits/rejected": -2.3397979736328125,
|
| 152 |
+
"logps/chosen": -267.89422607421875,
|
| 153 |
+
"logps/rejected": -286.52838134765625,
|
| 154 |
+
"loss": 0.6327,
|
| 155 |
+
"rewards/accuracies": 0.6875,
|
| 156 |
+
"rewards/chosen": -0.1534612625837326,
|
| 157 |
+
"rewards/margins": 0.16493618488311768,
|
| 158 |
+
"rewards/rejected": -0.3183974325656891,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.10468463752944256,
|
| 163 |
+
"grad_norm": 2.20348278766923,
|
| 164 |
+
"learning_rate": 4.9997324926814375e-06,
|
| 165 |
+
"logits/chosen": -2.408754587173462,
|
| 166 |
+
"logits/rejected": -2.332271099090576,
|
| 167 |
+
"logps/chosen": -306.8832702636719,
|
| 168 |
+
"logps/rejected": -322.60369873046875,
|
| 169 |
+
"loss": 0.6185,
|
| 170 |
+
"rewards/accuracies": 0.731249988079071,
|
| 171 |
+
"rewards/chosen": -0.3260679841041565,
|
| 172 |
+
"rewards/margins": 0.21181587874889374,
|
| 173 |
+
"rewards/rejected": -0.5378838777542114,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.10468463752944256,
|
| 178 |
+
"eval_logits/chosen": -2.3536956310272217,
|
| 179 |
+
"eval_logits/rejected": -2.295363426208496,
|
| 180 |
+
"eval_logps/chosen": -296.3735656738281,
|
| 181 |
+
"eval_logps/rejected": -300.7996826171875,
|
| 182 |
+
"eval_loss": 0.6239820718765259,
|
| 183 |
+
"eval_rewards/accuracies": 0.6964285969734192,
|
| 184 |
+
"eval_rewards/chosen": -0.30099427700042725,
|
| 185 |
+
"eval_rewards/margins": 0.23865534365177155,
|
| 186 |
+
"eval_rewards/rejected": -0.5396496653556824,
|
| 187 |
+
"eval_runtime": 276.8422,
|
| 188 |
+
"eval_samples_per_second": 7.224,
|
| 189 |
+
"eval_steps_per_second": 0.228,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.11515310128238682,
|
| 194 |
+
"grad_norm": 2.4992668200294634,
|
| 195 |
+
"learning_rate": 4.996723692767927e-06,
|
| 196 |
+
"logits/chosen": -2.4233577251434326,
|
| 197 |
+
"logits/rejected": -2.3276515007019043,
|
| 198 |
+
"logps/chosen": -282.03070068359375,
|
| 199 |
+
"logps/rejected": -270.5752258300781,
|
| 200 |
+
"loss": 0.6214,
|
| 201 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 202 |
+
"rewards/chosen": -0.32758888602256775,
|
| 203 |
+
"rewards/margins": 0.2726225256919861,
|
| 204 |
+
"rewards/rejected": -0.6002114415168762,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.12562156503533106,
|
| 209 |
+
"grad_norm": 2.277765435939027,
|
| 210 |
+
"learning_rate": 4.9903757462135984e-06,
|
| 211 |
+
"logits/chosen": -2.402243137359619,
|
| 212 |
+
"logits/rejected": -2.3531296253204346,
|
| 213 |
+
"logps/chosen": -279.3037414550781,
|
| 214 |
+
"logps/rejected": -320.41241455078125,
|
| 215 |
+
"loss": 0.5899,
|
| 216 |
+
"rewards/accuracies": 0.762499988079071,
|
| 217 |
+
"rewards/chosen": -0.3164636492729187,
|
| 218 |
+
"rewards/margins": 0.3019184470176697,
|
| 219 |
+
"rewards/rejected": -0.6183820962905884,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.1360900287882753,
|
| 224 |
+
"grad_norm": 3.5836859325532537,
|
| 225 |
+
"learning_rate": 4.980697142834315e-06,
|
| 226 |
+
"logits/chosen": -2.4281363487243652,
|
| 227 |
+
"logits/rejected": -2.3622257709503174,
|
| 228 |
+
"logps/chosen": -365.813232421875,
|
| 229 |
+
"logps/rejected": -333.8047790527344,
|
| 230 |
+
"loss": 0.5984,
|
| 231 |
+
"rewards/accuracies": 0.6875,
|
| 232 |
+
"rewards/chosen": -0.6453860402107239,
|
| 233 |
+
"rewards/margins": 0.291470468044281,
|
| 234 |
+
"rewards/rejected": -0.9368564486503601,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.14655849254121958,
|
| 239 |
+
"grad_norm": 4.177122005020008,
|
| 240 |
+
"learning_rate": 4.967700826904229e-06,
|
| 241 |
+
"logits/chosen": -2.3888461589813232,
|
| 242 |
+
"logits/rejected": -2.3509252071380615,
|
| 243 |
+
"logps/chosen": -329.9725341796875,
|
| 244 |
+
"logps/rejected": -369.2422790527344,
|
| 245 |
+
"loss": 0.5786,
|
| 246 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 247 |
+
"rewards/chosen": -0.5841600298881531,
|
| 248 |
+
"rewards/margins": 0.4313054084777832,
|
| 249 |
+
"rewards/rejected": -1.015465497970581,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.15702695629416383,
|
| 254 |
+
"grad_norm": 3.8223984872865686,
|
| 255 |
+
"learning_rate": 4.951404179843963e-06,
|
| 256 |
+
"logits/chosen": -2.4817895889282227,
|
| 257 |
+
"logits/rejected": -2.4037411212921143,
|
| 258 |
+
"logps/chosen": -363.9546203613281,
|
| 259 |
+
"logps/rejected": -355.483154296875,
|
| 260 |
+
"loss": 0.5982,
|
| 261 |
+
"rewards/accuracies": 0.699999988079071,
|
| 262 |
+
"rewards/chosen": -0.738756000995636,
|
| 263 |
+
"rewards/margins": 0.38375917077064514,
|
| 264 |
+
"rewards/rejected": -1.122515320777893,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.16749542004710807,
|
| 269 |
+
"grad_norm": 3.3391629933554023,
|
| 270 |
+
"learning_rate": 4.931828996974498e-06,
|
| 271 |
+
"logits/chosen": -2.4580092430114746,
|
| 272 |
+
"logits/rejected": -2.364123821258545,
|
| 273 |
+
"logps/chosen": -374.60455322265625,
|
| 274 |
+
"logps/rejected": -352.8321533203125,
|
| 275 |
+
"loss": 0.5851,
|
| 276 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 277 |
+
"rewards/chosen": -0.8589010238647461,
|
| 278 |
+
"rewards/margins": 0.3599582612514496,
|
| 279 |
+
"rewards/rejected": -1.2188594341278076,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.17796388380005235,
|
| 284 |
+
"grad_norm": 2.5964425242978337,
|
| 285 |
+
"learning_rate": 4.909001458367867e-06,
|
| 286 |
+
"logits/chosen": -2.4445998668670654,
|
| 287 |
+
"logits/rejected": -2.355095148086548,
|
| 288 |
+
"logps/chosen": -318.6025085449219,
|
| 289 |
+
"logps/rejected": -344.8854675292969,
|
| 290 |
+
"loss": 0.575,
|
| 291 |
+
"rewards/accuracies": 0.6875,
|
| 292 |
+
"rewards/chosen": -0.580207347869873,
|
| 293 |
+
"rewards/margins": 0.40731239318847656,
|
| 294 |
+
"rewards/rejected": -0.9875197410583496,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.1884323475529966,
|
| 299 |
+
"grad_norm": 3.6928906953423732,
|
| 300 |
+
"learning_rate": 4.882952093833628e-06,
|
| 301 |
+
"logits/chosen": -2.388700008392334,
|
| 302 |
+
"logits/rejected": -2.3635334968566895,
|
| 303 |
+
"logps/chosen": -325.55902099609375,
|
| 304 |
+
"logps/rejected": -366.1028747558594,
|
| 305 |
+
"loss": 0.55,
|
| 306 |
+
"rewards/accuracies": 0.6875,
|
| 307 |
+
"rewards/chosen": -0.6206040382385254,
|
| 308 |
+
"rewards/margins": 0.4654437005519867,
|
| 309 |
+
"rewards/rejected": -1.086047649383545,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.19890081130594087,
|
| 314 |
+
"grad_norm": 3.022202327145697,
|
| 315 |
+
"learning_rate": 4.853715742087947e-06,
|
| 316 |
+
"logits/chosen": -2.243128776550293,
|
| 317 |
+
"logits/rejected": -2.2242884635925293,
|
| 318 |
+
"logps/chosen": -345.83331298828125,
|
| 319 |
+
"logps/rejected": -424.47003173828125,
|
| 320 |
+
"loss": 0.564,
|
| 321 |
+
"rewards/accuracies": 0.793749988079071,
|
| 322 |
+
"rewards/chosen": -0.9772090911865234,
|
| 323 |
+
"rewards/margins": 0.5893430113792419,
|
| 324 |
+
"rewards/rejected": -1.5665521621704102,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.2093692750588851,
|
| 329 |
+
"grad_norm": 4.352581590749502,
|
| 330 |
+
"learning_rate": 4.821331504159906e-06,
|
| 331 |
+
"logits/chosen": -2.3912417888641357,
|
| 332 |
+
"logits/rejected": -2.310708999633789,
|
| 333 |
+
"logps/chosen": -394.71142578125,
|
| 334 |
+
"logps/rejected": -382.6004943847656,
|
| 335 |
+
"loss": 0.5724,
|
| 336 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 337 |
+
"rewards/chosen": -0.8045751452445984,
|
| 338 |
+
"rewards/margins": 0.4971863627433777,
|
| 339 |
+
"rewards/rejected": -1.3017616271972656,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.2093692750588851,
|
| 344 |
+
"eval_logits/chosen": -2.2930498123168945,
|
| 345 |
+
"eval_logits/rejected": -2.2448384761810303,
|
| 346 |
+
"eval_logps/chosen": -350.611328125,
|
| 347 |
+
"eval_logps/rejected": -379.675048828125,
|
| 348 |
+
"eval_loss": 0.5691524744033813,
|
| 349 |
+
"eval_rewards/accuracies": 0.7301587462425232,
|
| 350 |
+
"eval_rewards/chosen": -0.8433722257614136,
|
| 351 |
+
"eval_rewards/margins": 0.48503097891807556,
|
| 352 |
+
"eval_rewards/rejected": -1.3284029960632324,
|
| 353 |
+
"eval_runtime": 277.4907,
|
| 354 |
+
"eval_samples_per_second": 7.207,
|
| 355 |
+
"eval_steps_per_second": 0.227,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.21983773881182936,
|
| 360 |
+
"grad_norm": 4.349505936455232,
|
| 361 |
+
"learning_rate": 4.7858426910973435e-06,
|
| 362 |
+
"logits/chosen": -2.3390839099884033,
|
| 363 |
+
"logits/rejected": -2.296893835067749,
|
| 364 |
+
"logps/chosen": -365.0033264160156,
|
| 365 |
+
"logps/rejected": -390.9748229980469,
|
| 366 |
+
"loss": 0.5875,
|
| 367 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 368 |
+
"rewards/chosen": -0.8617814779281616,
|
| 369 |
+
"rewards/margins": 0.34669047594070435,
|
| 370 |
+
"rewards/rejected": -1.2084718942642212,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.23030620256477363,
|
| 375 |
+
"grad_norm": 3.456252004286951,
|
| 376 |
+
"learning_rate": 4.747296766042161e-06,
|
| 377 |
+
"logits/chosen": -2.345686435699463,
|
| 378 |
+
"logits/rejected": -2.2922120094299316,
|
| 379 |
+
"logps/chosen": -383.15887451171875,
|
| 380 |
+
"logps/rejected": -390.80059814453125,
|
| 381 |
+
"loss": 0.5635,
|
| 382 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 383 |
+
"rewards/chosen": -0.8896678686141968,
|
| 384 |
+
"rewards/margins": 0.5066917538642883,
|
| 385 |
+
"rewards/rejected": -1.3963596820831299,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.24077466631771788,
|
| 390 |
+
"grad_norm": 4.05851608400099,
|
| 391 |
+
"learning_rate": 4.705745280752586e-06,
|
| 392 |
+
"logits/chosen": -2.1309893131256104,
|
| 393 |
+
"logits/rejected": -2.105212450027466,
|
| 394 |
+
"logps/chosen": -367.39385986328125,
|
| 395 |
+
"logps/rejected": -395.93389892578125,
|
| 396 |
+
"loss": 0.5581,
|
| 397 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 398 |
+
"rewards/chosen": -1.184787631034851,
|
| 399 |
+
"rewards/margins": 0.43892526626586914,
|
| 400 |
+
"rewards/rejected": -1.6237128973007202,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.2512431300706621,
|
| 405 |
+
"grad_norm": 3.8267051093451507,
|
| 406 |
+
"learning_rate": 4.661243806657256e-06,
|
| 407 |
+
"logits/chosen": -2.1996395587921143,
|
| 408 |
+
"logits/rejected": -2.182788133621216,
|
| 409 |
+
"logps/chosen": -372.97308349609375,
|
| 410 |
+
"logps/rejected": -395.46978759765625,
|
| 411 |
+
"loss": 0.5428,
|
| 412 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 413 |
+
"rewards/chosen": -1.135464072227478,
|
| 414 |
+
"rewards/margins": 0.5388585925102234,
|
| 415 |
+
"rewards/rejected": -1.6743228435516357,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.26171159382360637,
|
| 420 |
+
"grad_norm": 4.611796368437759,
|
| 421 |
+
"learning_rate": 4.613851860533367e-06,
|
| 422 |
+
"logits/chosen": -2.163637399673462,
|
| 423 |
+
"logits/rejected": -2.143691062927246,
|
| 424 |
+
"logps/chosen": -375.72900390625,
|
| 425 |
+
"logps/rejected": -394.5628967285156,
|
| 426 |
+
"loss": 0.5294,
|
| 427 |
+
"rewards/accuracies": 0.75,
|
| 428 |
+
"rewards/chosen": -1.087786316871643,
|
| 429 |
+
"rewards/margins": 0.5885697603225708,
|
| 430 |
+
"rewards/rejected": -1.6763559579849243,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.2721800575765506,
|
| 435 |
+
"grad_norm": 4.723757696740007,
|
| 436 |
+
"learning_rate": 4.563632824908252e-06,
|
| 437 |
+
"logits/chosen": -2.07501482963562,
|
| 438 |
+
"logits/rejected": -2.003157138824463,
|
| 439 |
+
"logps/chosen": -431.285888671875,
|
| 440 |
+
"logps/rejected": -481.96435546875,
|
| 441 |
+
"loss": 0.536,
|
| 442 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 443 |
+
"rewards/chosen": -1.5983136892318726,
|
| 444 |
+
"rewards/margins": 0.6496692299842834,
|
| 445 |
+
"rewards/rejected": -2.247982978820801,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.2826485213294949,
|
| 450 |
+
"grad_norm": 6.406650480112877,
|
| 451 |
+
"learning_rate": 4.510653863290871e-06,
|
| 452 |
+
"logits/chosen": -1.9036792516708374,
|
| 453 |
+
"logits/rejected": -1.8629169464111328,
|
| 454 |
+
"logps/chosen": -441.50677490234375,
|
| 455 |
+
"logps/rejected": -505.6388244628906,
|
| 456 |
+
"loss": 0.5138,
|
| 457 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 458 |
+
"rewards/chosen": -1.6636412143707275,
|
| 459 |
+
"rewards/margins": 0.8482675552368164,
|
| 460 |
+
"rewards/rejected": -2.511908769607544,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.29311698508243916,
|
| 465 |
+
"grad_norm": 3.669198365138619,
|
| 466 |
+
"learning_rate": 4.454985830346574e-06,
|
| 467 |
+
"logits/chosen": -1.8643271923065186,
|
| 468 |
+
"logits/rejected": -1.7986135482788086,
|
| 469 |
+
"logps/chosen": -439.7417907714844,
|
| 470 |
+
"logps/rejected": -483.3232421875,
|
| 471 |
+
"loss": 0.5515,
|
| 472 |
+
"rewards/accuracies": 0.800000011920929,
|
| 473 |
+
"rewards/chosen": -1.501157522201538,
|
| 474 |
+
"rewards/margins": 0.768980085849762,
|
| 475 |
+
"rewards/rejected": -2.270137310028076,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.3035854488353834,
|
| 480 |
+
"grad_norm": 4.724976933585983,
|
| 481 |
+
"learning_rate": 4.396703177135262e-06,
|
| 482 |
+
"logits/chosen": -1.8855762481689453,
|
| 483 |
+
"logits/rejected": -1.7720863819122314,
|
| 484 |
+
"logps/chosen": -438.58154296875,
|
| 485 |
+
"logps/rejected": -466.4124450683594,
|
| 486 |
+
"loss": 0.5336,
|
| 487 |
+
"rewards/accuracies": 0.731249988079071,
|
| 488 |
+
"rewards/chosen": -1.535830020904541,
|
| 489 |
+
"rewards/margins": 0.6200370788574219,
|
| 490 |
+
"rewards/rejected": -2.155867099761963,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.31405391258832765,
|
| 495 |
+
"grad_norm": 4.575595350067731,
|
| 496 |
+
"learning_rate": 4.335883851539693e-06,
|
| 497 |
+
"logits/chosen": -1.7169132232666016,
|
| 498 |
+
"logits/rejected": -1.7286903858184814,
|
| 499 |
+
"logps/chosen": -402.37982177734375,
|
| 500 |
+
"logps/rejected": -462.84478759765625,
|
| 501 |
+
"loss": 0.5366,
|
| 502 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 503 |
+
"rewards/chosen": -1.5827662944793701,
|
| 504 |
+
"rewards/margins": 0.6053536534309387,
|
| 505 |
+
"rewards/rejected": -2.188120126724243,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.31405391258832765,
|
| 510 |
+
"eval_logits/chosen": -1.6850438117980957,
|
| 511 |
+
"eval_logits/rejected": -1.6220176219940186,
|
| 512 |
+
"eval_logps/chosen": -435.1428527832031,
|
| 513 |
+
"eval_logps/rejected": -495.46484375,
|
| 514 |
+
"eval_loss": 0.524876594543457,
|
| 515 |
+
"eval_rewards/accuracies": 0.7638888955116272,
|
| 516 |
+
"eval_rewards/chosen": -1.6886873245239258,
|
| 517 |
+
"eval_rewards/margins": 0.7976137399673462,
|
| 518 |
+
"eval_rewards/rejected": -2.4863009452819824,
|
| 519 |
+
"eval_runtime": 278.1285,
|
| 520 |
+
"eval_samples_per_second": 7.191,
|
| 521 |
+
"eval_steps_per_second": 0.227,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.3245223763412719,
|
| 526 |
+
"grad_norm": 4.624400922286726,
|
| 527 |
+
"learning_rate": 4.2726091940171055e-06,
|
| 528 |
+
"logits/chosen": -1.75604248046875,
|
| 529 |
+
"logits/rejected": -1.6548255681991577,
|
| 530 |
+
"logps/chosen": -441.04571533203125,
|
| 531 |
+
"logps/rejected": -495.0343322753906,
|
| 532 |
+
"loss": 0.5239,
|
| 533 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 534 |
+
"rewards/chosen": -1.6734451055526733,
|
| 535 |
+
"rewards/margins": 0.911232590675354,
|
| 536 |
+
"rewards/rejected": -2.5846774578094482,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.33499084009421615,
|
| 541 |
+
"grad_norm": 5.321599244496586,
|
| 542 |
+
"learning_rate": 4.206963828813555e-06,
|
| 543 |
+
"logits/chosen": -1.7151927947998047,
|
| 544 |
+
"logits/rejected": -1.659000039100647,
|
| 545 |
+
"logps/chosen": -407.6772766113281,
|
| 546 |
+
"logps/rejected": -487.42816162109375,
|
| 547 |
+
"loss": 0.5533,
|
| 548 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 549 |
+
"rewards/chosen": -1.6879345178604126,
|
| 550 |
+
"rewards/margins": 0.7429603338241577,
|
| 551 |
+
"rewards/rejected": -2.4308950901031494,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.34545930384716045,
|
| 556 |
+
"grad_norm": 4.6340466714580755,
|
| 557 |
+
"learning_rate": 4.139035550786495e-06,
|
| 558 |
+
"logits/chosen": -1.693902611732483,
|
| 559 |
+
"logits/rejected": -1.6580374240875244,
|
| 560 |
+
"logps/chosen": -437.0306091308594,
|
| 561 |
+
"logps/rejected": -501.24298095703125,
|
| 562 |
+
"loss": 0.5172,
|
| 563 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 564 |
+
"rewards/chosen": -1.860239028930664,
|
| 565 |
+
"rewards/margins": 0.6861482858657837,
|
| 566 |
+
"rewards/rejected": -2.5463874340057373,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.3559277676001047,
|
| 571 |
+
"grad_norm": 4.529361319396784,
|
| 572 |
+
"learning_rate": 4.068915207986931e-06,
|
| 573 |
+
"logits/chosen": -1.5517462491989136,
|
| 574 |
+
"logits/rejected": -1.497995376586914,
|
| 575 |
+
"logps/chosen": -427.9058532714844,
|
| 576 |
+
"logps/rejected": -476.1865234375,
|
| 577 |
+
"loss": 0.5503,
|
| 578 |
+
"rewards/accuracies": 0.699999988079071,
|
| 579 |
+
"rewards/chosen": -1.7529428005218506,
|
| 580 |
+
"rewards/margins": 0.7496469020843506,
|
| 581 |
+
"rewards/rejected": -2.502589702606201,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.36639623135304894,
|
| 586 |
+
"grad_norm": 5.461159855322785,
|
| 587 |
+
"learning_rate": 3.996696580158211e-06,
|
| 588 |
+
"logits/chosen": -1.7160961627960205,
|
| 589 |
+
"logits/rejected": -1.64591383934021,
|
| 590 |
+
"logps/chosen": -440.3072204589844,
|
| 591 |
+
"logps/rejected": -496.3219299316406,
|
| 592 |
+
"loss": 0.5145,
|
| 593 |
+
"rewards/accuracies": 0.668749988079071,
|
| 594 |
+
"rewards/chosen": -1.8336540460586548,
|
| 595 |
+
"rewards/margins": 0.7150315046310425,
|
| 596 |
+
"rewards/rejected": -2.5486855506896973,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.3768646951059932,
|
| 601 |
+
"grad_norm": 5.022405804138076,
|
| 602 |
+
"learning_rate": 3.922476253313921e-06,
|
| 603 |
+
"logits/chosen": -1.6719996929168701,
|
| 604 |
+
"logits/rejected": -1.6425079107284546,
|
| 605 |
+
"logps/chosen": -405.40142822265625,
|
| 606 |
+
"logps/rejected": -469.779052734375,
|
| 607 |
+
"loss": 0.4736,
|
| 608 |
+
"rewards/accuracies": 0.800000011920929,
|
| 609 |
+
"rewards/chosen": -1.4964035749435425,
|
| 610 |
+
"rewards/margins": 0.8178457021713257,
|
| 611 |
+
"rewards/rejected": -2.314249277114868,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.38733315885893743,
|
| 616 |
+
"grad_norm": 4.765071999966862,
|
| 617 |
+
"learning_rate": 3.846353490562664e-06,
|
| 618 |
+
"logits/chosen": -1.754809021949768,
|
| 619 |
+
"logits/rejected": -1.7150640487670898,
|
| 620 |
+
"logps/chosen": -381.4289855957031,
|
| 621 |
+
"logps/rejected": -456.10675048828125,
|
| 622 |
+
"loss": 0.5203,
|
| 623 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 624 |
+
"rewards/chosen": -1.3520570993423462,
|
| 625 |
+
"rewards/margins": 0.8618815541267395,
|
| 626 |
+
"rewards/rejected": -2.2139384746551514,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.39780162261188173,
|
| 631 |
+
"grad_norm": 5.077040951702186,
|
| 632 |
+
"learning_rate": 3.768430099352445e-06,
|
| 633 |
+
"logits/chosen": -1.7217756509780884,
|
| 634 |
+
"logits/rejected": -1.67500901222229,
|
| 635 |
+
"logps/chosen": -441.7967834472656,
|
| 636 |
+
"logps/rejected": -520.3387451171875,
|
| 637 |
+
"loss": 0.5364,
|
| 638 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 639 |
+
"rewards/chosen": -1.86903977394104,
|
| 640 |
+
"rewards/margins": 0.7624320983886719,
|
| 641 |
+
"rewards/rejected": -2.631471872329712,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.408270086364826,
|
| 646 |
+
"grad_norm": 3.6370311779518514,
|
| 647 |
+
"learning_rate": 3.6888102953122307e-06,
|
| 648 |
+
"logits/chosen": -1.7671334743499756,
|
| 649 |
+
"logits/rejected": -1.7222200632095337,
|
| 650 |
+
"logps/chosen": -417.48699951171875,
|
| 651 |
+
"logps/rejected": -465.5294494628906,
|
| 652 |
+
"loss": 0.5336,
|
| 653 |
+
"rewards/accuracies": 0.6875,
|
| 654 |
+
"rewards/chosen": -1.583051323890686,
|
| 655 |
+
"rewards/margins": 0.6584590673446655,
|
| 656 |
+
"rewards/rejected": -2.2415101528167725,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.4187385501177702,
|
| 661 |
+
"grad_norm": 3.8429456015253396,
|
| 662 |
+
"learning_rate": 3.607600562872785e-06,
|
| 663 |
+
"logits/chosen": -1.843812346458435,
|
| 664 |
+
"logits/rejected": -1.8115432262420654,
|
| 665 |
+
"logps/chosen": -375.5889587402344,
|
| 666 |
+
"logps/rejected": -427.83990478515625,
|
| 667 |
+
"loss": 0.5397,
|
| 668 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 669 |
+
"rewards/chosen": -1.298298954963684,
|
| 670 |
+
"rewards/margins": 0.5158218145370483,
|
| 671 |
+
"rewards/rejected": -1.8141206502914429,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.4187385501177702,
|
| 676 |
+
"eval_logits/chosen": -1.8144292831420898,
|
| 677 |
+
"eval_logits/rejected": -1.7585728168487549,
|
| 678 |
+
"eval_logps/chosen": -396.2536926269531,
|
| 679 |
+
"eval_logps/rejected": -446.0619201660156,
|
| 680 |
+
"eval_loss": 0.5252955555915833,
|
| 681 |
+
"eval_rewards/accuracies": 0.7698412537574768,
|
| 682 |
+
"eval_rewards/chosen": -1.2997959852218628,
|
| 683 |
+
"eval_rewards/margins": 0.6924758553504944,
|
| 684 |
+
"eval_rewards/rejected": -1.9922715425491333,
|
| 685 |
+
"eval_runtime": 277.0739,
|
| 686 |
+
"eval_samples_per_second": 7.218,
|
| 687 |
+
"eval_steps_per_second": 0.227,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.42920701387071447,
|
| 692 |
+
"grad_norm": 4.321854918430961,
|
| 693 |
+
"learning_rate": 3.5249095128531863e-06,
|
| 694 |
+
"logits/chosen": -1.708707571029663,
|
| 695 |
+
"logits/rejected": -1.6425899267196655,
|
| 696 |
+
"logps/chosen": -434.7601623535156,
|
| 697 |
+
"logps/rejected": -490.0912170410156,
|
| 698 |
+
"loss": 0.5406,
|
| 699 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 700 |
+
"rewards/chosen": -1.66852605342865,
|
| 701 |
+
"rewards/margins": 0.8226760625839233,
|
| 702 |
+
"rewards/rejected": -2.4912023544311523,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.4396754776236587,
|
| 707 |
+
"grad_norm": 4.6429659320397185,
|
| 708 |
+
"learning_rate": 3.4408477372034743e-06,
|
| 709 |
+
"logits/chosen": -1.7646077871322632,
|
| 710 |
+
"logits/rejected": -1.6311779022216797,
|
| 711 |
+
"logps/chosen": -465.91851806640625,
|
| 712 |
+
"logps/rejected": -482.90313720703125,
|
| 713 |
+
"loss": 0.5232,
|
| 714 |
+
"rewards/accuracies": 0.706250011920929,
|
| 715 |
+
"rewards/chosen": -1.8112109899520874,
|
| 716 |
+
"rewards/margins": 0.7522093653678894,
|
| 717 |
+
"rewards/rejected": -2.5634207725524902,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.45014394137660296,
|
| 722 |
+
"grad_norm": 4.073383292817544,
|
| 723 |
+
"learning_rate": 3.355527661097728e-06,
|
| 724 |
+
"logits/chosen": -1.7000205516815186,
|
| 725 |
+
"logits/rejected": -1.6602411270141602,
|
| 726 |
+
"logps/chosen": -406.20111083984375,
|
| 727 |
+
"logps/rejected": -491.0992126464844,
|
| 728 |
+
"loss": 0.5113,
|
| 729 |
+
"rewards/accuracies": 0.75,
|
| 730 |
+
"rewards/chosen": -1.771209478378296,
|
| 731 |
+
"rewards/margins": 0.8097410202026367,
|
| 732 |
+
"rewards/rejected": -2.5809504985809326,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.46061240512954726,
|
| 737 |
+
"grad_norm": 6.363547319828774,
|
| 738 |
+
"learning_rate": 3.269063392575352e-06,
|
| 739 |
+
"logits/chosen": -1.6135189533233643,
|
| 740 |
+
"logits/rejected": -1.6108068227767944,
|
| 741 |
+
"logps/chosen": -434.0970764160156,
|
| 742 |
+
"logps/rejected": -488.9344787597656,
|
| 743 |
+
"loss": 0.5193,
|
| 744 |
+
"rewards/accuracies": 0.706250011920929,
|
| 745 |
+
"rewards/chosen": -1.8390247821807861,
|
| 746 |
+
"rewards/margins": 0.7370558977127075,
|
| 747 |
+
"rewards/rejected": -2.576080799102783,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.4710808688824915,
|
| 752 |
+
"grad_norm": 5.120415362893221,
|
| 753 |
+
"learning_rate": 3.181570569931697e-06,
|
| 754 |
+
"logits/chosen": -1.699637770652771,
|
| 755 |
+
"logits/rejected": -1.6541831493377686,
|
| 756 |
+
"logps/chosen": -434.42010498046875,
|
| 757 |
+
"logps/rejected": -562.7218017578125,
|
| 758 |
+
"loss": 0.508,
|
| 759 |
+
"rewards/accuracies": 0.768750011920929,
|
| 760 |
+
"rewards/chosen": -1.8876171112060547,
|
| 761 |
+
"rewards/margins": 1.1184637546539307,
|
| 762 |
+
"rewards/rejected": -3.0060808658599854,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.48154933263543576,
|
| 767 |
+
"grad_norm": 4.704501376897649,
|
| 768 |
+
"learning_rate": 3.09316620706208e-06,
|
| 769 |
+
"logits/chosen": -1.7134917974472046,
|
| 770 |
+
"logits/rejected": -1.6925232410430908,
|
| 771 |
+
"logps/chosen": -452.0691833496094,
|
| 772 |
+
"logps/rejected": -523.4850463867188,
|
| 773 |
+
"loss": 0.5335,
|
| 774 |
+
"rewards/accuracies": 0.731249988079071,
|
| 775 |
+
"rewards/chosen": -1.904476523399353,
|
| 776 |
+
"rewards/margins": 0.8279927372932434,
|
| 777 |
+
"rewards/rejected": -2.732469320297241,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.49201779638838,
|
| 782 |
+
"grad_norm": 4.879700738283134,
|
| 783 |
+
"learning_rate": 3.0039685369660785e-06,
|
| 784 |
+
"logits/chosen": -1.7574955224990845,
|
| 785 |
+
"logits/rejected": -1.666815161705017,
|
| 786 |
+
"logps/chosen": -433.5028381347656,
|
| 787 |
+
"logps/rejected": -469.12158203125,
|
| 788 |
+
"loss": 0.4923,
|
| 789 |
+
"rewards/accuracies": 0.699999988079071,
|
| 790 |
+
"rewards/chosen": -1.7322635650634766,
|
| 791 |
+
"rewards/margins": 0.6690031290054321,
|
| 792 |
+
"rewards/rejected": -2.401266574859619,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.5024862601413242,
|
| 797 |
+
"grad_norm": 5.236704166005388,
|
| 798 |
+
"learning_rate": 2.91409685362137e-06,
|
| 799 |
+
"logits/chosen": -1.6420507431030273,
|
| 800 |
+
"logits/rejected": -1.551762342453003,
|
| 801 |
+
"logps/chosen": -483.396240234375,
|
| 802 |
+
"logps/rejected": -551.7774658203125,
|
| 803 |
+
"loss": 0.5032,
|
| 804 |
+
"rewards/accuracies": 0.800000011920929,
|
| 805 |
+
"rewards/chosen": -2.1131300926208496,
|
| 806 |
+
"rewards/margins": 0.9647353887557983,
|
| 807 |
+
"rewards/rejected": -3.0778656005859375,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.5129547238942685,
|
| 812 |
+
"grad_norm": 4.9932627461086625,
|
| 813 |
+
"learning_rate": 2.8236713524386085e-06,
|
| 814 |
+
"logits/chosen": -1.7080411911010742,
|
| 815 |
+
"logits/rejected": -1.606384515762329,
|
| 816 |
+
"logps/chosen": -494.674560546875,
|
| 817 |
+
"logps/rejected": -555.2131958007812,
|
| 818 |
+
"loss": 0.497,
|
| 819 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 820 |
+
"rewards/chosen": -2.097015619277954,
|
| 821 |
+
"rewards/margins": 0.9440878629684448,
|
| 822 |
+
"rewards/rejected": -3.0411033630371094,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.5234231876472127,
|
| 827 |
+
"grad_norm": 4.897787536727827,
|
| 828 |
+
"learning_rate": 2.7328129695107205e-06,
|
| 829 |
+
"logits/chosen": -1.6698167324066162,
|
| 830 |
+
"logits/rejected": -1.6173324584960938,
|
| 831 |
+
"logps/chosen": -458.3126525878906,
|
| 832 |
+
"logps/rejected": -532.6166381835938,
|
| 833 |
+
"loss": 0.5003,
|
| 834 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 835 |
+
"rewards/chosen": -1.9204323291778564,
|
| 836 |
+
"rewards/margins": 0.9951194524765015,
|
| 837 |
+
"rewards/rejected": -2.9155516624450684,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.5234231876472127,
|
| 842 |
+
"eval_logits/chosen": -1.6681629419326782,
|
| 843 |
+
"eval_logits/rejected": -1.6049078702926636,
|
| 844 |
+
"eval_logps/chosen": -466.0908508300781,
|
| 845 |
+
"eval_logps/rejected": -538.906494140625,
|
| 846 |
+
"eval_loss": 0.5013344287872314,
|
| 847 |
+
"eval_rewards/accuracies": 0.7658730149269104,
|
| 848 |
+
"eval_rewards/chosen": -1.9981670379638672,
|
| 849 |
+
"eval_rewards/margins": 0.922550618648529,
|
| 850 |
+
"eval_rewards/rejected": -2.92071795463562,
|
| 851 |
+
"eval_runtime": 276.7176,
|
| 852 |
+
"eval_samples_per_second": 7.228,
|
| 853 |
+
"eval_steps_per_second": 0.228,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.533891651400157,
|
| 858 |
+
"grad_norm": 4.97316270122908,
|
| 859 |
+
"learning_rate": 2.641643219871597e-06,
|
| 860 |
+
"logits/chosen": -1.7244354486465454,
|
| 861 |
+
"logits/rejected": -1.6957614421844482,
|
| 862 |
+
"logps/chosen": -451.49981689453125,
|
| 863 |
+
"logps/rejected": -516.1077270507812,
|
| 864 |
+
"loss": 0.4933,
|
| 865 |
+
"rewards/accuracies": 0.706250011920929,
|
| 866 |
+
"rewards/chosen": -1.911877989768982,
|
| 867 |
+
"rewards/margins": 0.8705556988716125,
|
| 868 |
+
"rewards/rejected": -2.7824337482452393,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.5443601151531012,
|
| 873 |
+
"grad_norm": 5.8930442835078685,
|
| 874 |
+
"learning_rate": 2.5502840349805074e-06,
|
| 875 |
+
"logits/chosen": -1.6949068307876587,
|
| 876 |
+
"logits/rejected": -1.6189374923706055,
|
| 877 |
+
"logps/chosen": -499.230224609375,
|
| 878 |
+
"logps/rejected": -536.48583984375,
|
| 879 |
+
"loss": 0.5261,
|
| 880 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 881 |
+
"rewards/chosen": -2.2237086296081543,
|
| 882 |
+
"rewards/margins": 0.827392578125,
|
| 883 |
+
"rewards/rejected": -3.0511016845703125,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.5548285789060455,
|
| 888 |
+
"grad_norm": 5.076166821402116,
|
| 889 |
+
"learning_rate": 2.4588575996495797e-06,
|
| 890 |
+
"logits/chosen": -1.6722028255462646,
|
| 891 |
+
"logits/rejected": -1.606205940246582,
|
| 892 |
+
"logps/chosen": -441.3253479003906,
|
| 893 |
+
"logps/rejected": -515.5338134765625,
|
| 894 |
+
"loss": 0.4997,
|
| 895 |
+
"rewards/accuracies": 0.824999988079071,
|
| 896 |
+
"rewards/chosen": -1.6732591390609741,
|
| 897 |
+
"rewards/margins": 0.9954363703727722,
|
| 898 |
+
"rewards/rejected": -2.6686952114105225,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.5652970426589898,
|
| 903 |
+
"grad_norm": 5.298568838364531,
|
| 904 |
+
"learning_rate": 2.367486188632446e-06,
|
| 905 |
+
"logits/chosen": -1.6246812343597412,
|
| 906 |
+
"logits/rejected": -1.564178705215454,
|
| 907 |
+
"logps/chosen": -443.343017578125,
|
| 908 |
+
"logps/rejected": -481.7394104003906,
|
| 909 |
+
"loss": 0.5271,
|
| 910 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 911 |
+
"rewards/chosen": -1.873234510421753,
|
| 912 |
+
"rewards/margins": 0.8055590391159058,
|
| 913 |
+
"rewards/rejected": -2.6787936687469482,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.575765506411934,
|
| 918 |
+
"grad_norm": 4.903563195070263,
|
| 919 |
+
"learning_rate": 2.276292003092593e-06,
|
| 920 |
+
"logits/chosen": -1.591386079788208,
|
| 921 |
+
"logits/rejected": -1.5199774503707886,
|
| 922 |
+
"logps/chosen": -461.0108337402344,
|
| 923 |
+
"logps/rejected": -490.60498046875,
|
| 924 |
+
"loss": 0.517,
|
| 925 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 926 |
+
"rewards/chosen": -1.9846340417861938,
|
| 927 |
+
"rewards/margins": 0.6464044451713562,
|
| 928 |
+
"rewards/rejected": -2.6310389041900635,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.5862339701648783,
|
| 933 |
+
"grad_norm": 6.151878065059773,
|
| 934 |
+
"learning_rate": 2.1853970071701415e-06,
|
| 935 |
+
"logits/chosen": -1.524752140045166,
|
| 936 |
+
"logits/rejected": -1.4780924320220947,
|
| 937 |
+
"logps/chosen": -458.1922912597656,
|
| 938 |
+
"logps/rejected": -509.72918701171875,
|
| 939 |
+
"loss": 0.5217,
|
| 940 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 941 |
+
"rewards/chosen": -1.9151818752288818,
|
| 942 |
+
"rewards/margins": 0.8629971742630005,
|
| 943 |
+
"rewards/rejected": -2.7781789302825928,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.5967024339178225,
|
| 948 |
+
"grad_norm": 5.220341094167877,
|
| 949 |
+
"learning_rate": 2.0949227648656194e-06,
|
| 950 |
+
"logits/chosen": -1.5827438831329346,
|
| 951 |
+
"logits/rejected": -1.5035909414291382,
|
| 952 |
+
"logps/chosen": -455.52197265625,
|
| 953 |
+
"logps/rejected": -528.5602416992188,
|
| 954 |
+
"loss": 0.5069,
|
| 955 |
+
"rewards/accuracies": 0.762499988079071,
|
| 956 |
+
"rewards/chosen": -1.9370582103729248,
|
| 957 |
+
"rewards/margins": 0.8514412641525269,
|
| 958 |
+
"rewards/rejected": -2.788499355316162,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.6071708976707668,
|
| 963 |
+
"grad_norm": 5.5150025763626145,
|
| 964 |
+
"learning_rate": 2.00499027745888e-06,
|
| 965 |
+
"logits/chosen": -1.5897815227508545,
|
| 966 |
+
"logits/rejected": -1.5100150108337402,
|
| 967 |
+
"logps/chosen": -404.41741943359375,
|
| 968 |
+
"logps/rejected": -460.427001953125,
|
| 969 |
+
"loss": 0.5166,
|
| 970 |
+
"rewards/accuracies": 0.737500011920929,
|
| 971 |
+
"rewards/chosen": -1.6832536458969116,
|
| 972 |
+
"rewards/margins": 0.7988817095756531,
|
| 973 |
+
"rewards/rejected": -2.48213529586792,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.6176393614237111,
|
| 978 |
+
"grad_norm": 4.9008082837332765,
|
| 979 |
+
"learning_rate": 1.915719821680624e-06,
|
| 980 |
+
"logits/chosen": -1.6151273250579834,
|
| 981 |
+
"logits/rejected": -1.5963796377182007,
|
| 982 |
+
"logps/chosen": -419.74725341796875,
|
| 983 |
+
"logps/rejected": -513.4180908203125,
|
| 984 |
+
"loss": 0.494,
|
| 985 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 986 |
+
"rewards/chosen": -1.7207034826278687,
|
| 987 |
+
"rewards/margins": 0.8950328826904297,
|
| 988 |
+
"rewards/rejected": -2.615736246109009,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.6281078251766553,
|
| 993 |
+
"grad_norm": 6.4962338338594385,
|
| 994 |
+
"learning_rate": 1.8272307888529276e-06,
|
| 995 |
+
"logits/chosen": -1.4385050535202026,
|
| 996 |
+
"logits/rejected": -1.3539499044418335,
|
| 997 |
+
"logps/chosen": -534.6634521484375,
|
| 998 |
+
"logps/rejected": -584.0506591796875,
|
| 999 |
+
"loss": 0.4835,
|
| 1000 |
+
"rewards/accuracies": 0.78125,
|
| 1001 |
+
"rewards/chosen": -2.359952449798584,
|
| 1002 |
+
"rewards/margins": 0.8336412310600281,
|
| 1003 |
+
"rewards/rejected": -3.193593740463257,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.6281078251766553,
|
| 1008 |
+
"eval_logits/chosen": -1.4124971628189087,
|
| 1009 |
+
"eval_logits/rejected": -1.3416856527328491,
|
| 1010 |
+
"eval_logps/chosen": -523.2593383789062,
|
| 1011 |
+
"eval_logps/rejected": -598.5182495117188,
|
| 1012 |
+
"eval_loss": 0.5026703476905823,
|
| 1013 |
+
"eval_rewards/accuracies": 0.7559523582458496,
|
| 1014 |
+
"eval_rewards/chosen": -2.569852352142334,
|
| 1015 |
+
"eval_rewards/margins": 0.9469824433326721,
|
| 1016 |
+
"eval_rewards/rejected": -3.516834259033203,
|
| 1017 |
+
"eval_runtime": 276.7094,
|
| 1018 |
+
"eval_samples_per_second": 7.228,
|
| 1019 |
+
"eval_steps_per_second": 0.228,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.6385762889295996,
|
| 1024 |
+
"grad_norm": 5.595781300544392,
|
| 1025 |
+
"learning_rate": 1.739641525213929e-06,
|
| 1026 |
+
"logits/chosen": -1.5086050033569336,
|
| 1027 |
+
"logits/rejected": -1.478257179260254,
|
| 1028 |
+
"logps/chosen": -520.480712890625,
|
| 1029 |
+
"logps/rejected": -586.702880859375,
|
| 1030 |
+
"loss": 0.5161,
|
| 1031 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1032 |
+
"rewards/chosen": -2.468291997909546,
|
| 1033 |
+
"rewards/margins": 1.007550835609436,
|
| 1034 |
+
"rewards/rejected": -3.4758429527282715,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.6490447526825438,
|
| 1039 |
+
"grad_norm": 6.142511293341612,
|
| 1040 |
+
"learning_rate": 1.6530691736402317e-06,
|
| 1041 |
+
"logits/chosen": -1.6536248922348022,
|
| 1042 |
+
"logits/rejected": -1.5585447549819946,
|
| 1043 |
+
"logps/chosen": -506.4708557128906,
|
| 1044 |
+
"logps/rejected": -550.4331665039062,
|
| 1045 |
+
"loss": 0.4796,
|
| 1046 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1047 |
+
"rewards/chosen": -2.0505268573760986,
|
| 1048 |
+
"rewards/margins": 0.9777840375900269,
|
| 1049 |
+
"rewards/rejected": -3.028310775756836,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.6595132164354881,
|
| 1054 |
+
"grad_norm": 7.522645087833883,
|
| 1055 |
+
"learning_rate": 1.5676295169786864e-06,
|
| 1056 |
+
"logits/chosen": -1.6819908618927002,
|
| 1057 |
+
"logits/rejected": -1.6001083850860596,
|
| 1058 |
+
"logps/chosen": -470.88494873046875,
|
| 1059 |
+
"logps/rejected": -511.28680419921875,
|
| 1060 |
+
"loss": 0.4863,
|
| 1061 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1062 |
+
"rewards/chosen": -1.9194360971450806,
|
| 1063 |
+
"rewards/margins": 0.8219398260116577,
|
| 1064 |
+
"rewards/rejected": -2.741375684738159,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.6699816801884323,
|
| 1069 |
+
"grad_norm": 7.853260652377011,
|
| 1070 |
+
"learning_rate": 1.4834368231970922e-06,
|
| 1071 |
+
"logits/chosen": -1.564366340637207,
|
| 1072 |
+
"logits/rejected": -1.494086742401123,
|
| 1073 |
+
"logps/chosen": -437.3993225097656,
|
| 1074 |
+
"logps/rejected": -517.3038330078125,
|
| 1075 |
+
"loss": 0.4786,
|
| 1076 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1077 |
+
"rewards/chosen": -1.9992519617080688,
|
| 1078 |
+
"rewards/margins": 0.9533640742301941,
|
| 1079 |
+
"rewards/rejected": -2.952615976333618,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.6804501439413766,
|
| 1084 |
+
"grad_norm": 5.803651393914848,
|
| 1085 |
+
"learning_rate": 1.4006036925609245e-06,
|
| 1086 |
+
"logits/chosen": -1.6179273128509521,
|
| 1087 |
+
"logits/rejected": -1.5984830856323242,
|
| 1088 |
+
"logps/chosen": -457.8565979003906,
|
| 1089 |
+
"logps/rejected": -540.9107666015625,
|
| 1090 |
+
"loss": 0.5039,
|
| 1091 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1092 |
+
"rewards/chosen": -1.9681861400604248,
|
| 1093 |
+
"rewards/margins": 0.8296257257461548,
|
| 1094 |
+
"rewards/rejected": -2.79781174659729,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6909186076943209,
|
| 1099 |
+
"grad_norm": 5.141240530191721,
|
| 1100 |
+
"learning_rate": 1.3192409070404582e-06,
|
| 1101 |
+
"logits/chosen": -1.680153489112854,
|
| 1102 |
+
"logits/rejected": -1.6061805486679077,
|
| 1103 |
+
"logps/chosen": -439.73272705078125,
|
| 1104 |
+
"logps/rejected": -508.1709899902344,
|
| 1105 |
+
"loss": 0.4981,
|
| 1106 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1107 |
+
"rewards/chosen": -1.8612697124481201,
|
| 1108 |
+
"rewards/margins": 0.8554447889328003,
|
| 1109 |
+
"rewards/rejected": -2.716714382171631,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.7013870714472651,
|
| 1114 |
+
"grad_norm": 6.25377182984675,
|
| 1115 |
+
"learning_rate": 1.2394572821496953e-06,
|
| 1116 |
+
"logits/chosen": -1.5166045427322388,
|
| 1117 |
+
"logits/rejected": -1.4167017936706543,
|
| 1118 |
+
"logps/chosen": -456.9510192871094,
|
| 1119 |
+
"logps/rejected": -535.3768310546875,
|
| 1120 |
+
"loss": 0.4811,
|
| 1121 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1122 |
+
"rewards/chosen": -2.0087170600891113,
|
| 1123 |
+
"rewards/margins": 1.0361051559448242,
|
| 1124 |
+
"rewards/rejected": -3.0448222160339355,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.7118555352002094,
|
| 1129 |
+
"grad_norm": 7.572206476060471,
|
| 1130 |
+
"learning_rate": 1.1613595214152713e-06,
|
| 1131 |
+
"logits/chosen": -1.5004642009735107,
|
| 1132 |
+
"logits/rejected": -1.4637364149093628,
|
| 1133 |
+
"logps/chosen": -480.53875732421875,
|
| 1134 |
+
"logps/rejected": -548.7276611328125,
|
| 1135 |
+
"loss": 0.5205,
|
| 1136 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1137 |
+
"rewards/chosen": -2.131697416305542,
|
| 1138 |
+
"rewards/margins": 0.8276796340942383,
|
| 1139 |
+
"rewards/rejected": -2.9593770503997803,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.7223239989531536,
|
| 1144 |
+
"grad_norm": 4.010350785789718,
|
| 1145 |
+
"learning_rate": 1.0850520736699362e-06,
|
| 1146 |
+
"logits/chosen": -1.588531255722046,
|
| 1147 |
+
"logits/rejected": -1.503158688545227,
|
| 1148 |
+
"logps/chosen": -480.64715576171875,
|
| 1149 |
+
"logps/rejected": -513.88232421875,
|
| 1150 |
+
"loss": 0.4544,
|
| 1151 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1152 |
+
"rewards/chosen": -2.0653069019317627,
|
| 1153 |
+
"rewards/margins": 0.8280428647994995,
|
| 1154 |
+
"rewards/rejected": -2.8933494091033936,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.7327924627060979,
|
| 1159 |
+
"grad_norm": 5.214297382993777,
|
| 1160 |
+
"learning_rate": 1.0106369933615043e-06,
|
| 1161 |
+
"logits/chosen": -1.4556739330291748,
|
| 1162 |
+
"logits/rejected": -1.4368045330047607,
|
| 1163 |
+
"logps/chosen": -476.2962951660156,
|
| 1164 |
+
"logps/rejected": -560.508056640625,
|
| 1165 |
+
"loss": 0.4715,
|
| 1166 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1167 |
+
"rewards/chosen": -2.2485156059265137,
|
| 1168 |
+
"rewards/margins": 0.9335169792175293,
|
| 1169 |
+
"rewards/rejected": -3.182032585144043,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.7327924627060979,
|
| 1174 |
+
"eval_logits/chosen": -1.4479634761810303,
|
| 1175 |
+
"eval_logits/rejected": -1.378204107284546,
|
| 1176 |
+
"eval_logps/chosen": -485.28936767578125,
|
| 1177 |
+
"eval_logps/rejected": -566.1954956054688,
|
| 1178 |
+
"eval_loss": 0.4955996572971344,
|
| 1179 |
+
"eval_rewards/accuracies": 0.7678571343421936,
|
| 1180 |
+
"eval_rewards/chosen": -2.190152883529663,
|
| 1181 |
+
"eval_rewards/margins": 1.0034549236297607,
|
| 1182 |
+
"eval_rewards/rejected": -3.193607807159424,
|
| 1183 |
+
"eval_runtime": 277.1042,
|
| 1184 |
+
"eval_samples_per_second": 7.218,
|
| 1185 |
+
"eval_steps_per_second": 0.227,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.7432609264590422,
|
| 1190 |
+
"grad_norm": 5.527625380225802,
|
| 1191 |
+
"learning_rate": 9.382138040640714e-07,
|
| 1192 |
+
"logits/chosen": -1.4906096458435059,
|
| 1193 |
+
"logits/rejected": -1.3870160579681396,
|
| 1194 |
+
"logps/chosen": -493.5198669433594,
|
| 1195 |
+
"logps/rejected": -554.8124389648438,
|
| 1196 |
+
"loss": 0.5528,
|
| 1197 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1198 |
+
"rewards/chosen": -2.260972499847412,
|
| 1199 |
+
"rewards/margins": 0.9154504537582397,
|
| 1200 |
+
"rewards/rejected": -3.176422595977783,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.7537293902119864,
|
| 1205 |
+
"grad_norm": 4.7649826972694145,
|
| 1206 |
+
"learning_rate": 8.678793653740633e-07,
|
| 1207 |
+
"logits/chosen": -1.6233956813812256,
|
| 1208 |
+
"logits/rejected": -1.5308845043182373,
|
| 1209 |
+
"logps/chosen": -514.6124267578125,
|
| 1210 |
+
"logps/rejected": -569.4010620117188,
|
| 1211 |
+
"loss": 0.4833,
|
| 1212 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1213 |
+
"rewards/chosen": -1.9400179386138916,
|
| 1214 |
+
"rewards/margins": 1.038830280303955,
|
| 1215 |
+
"rewards/rejected": -2.9788479804992676,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.7641978539649307,
|
| 1220 |
+
"grad_norm": 5.43475547987397,
|
| 1221 |
+
"learning_rate": 7.997277433690984e-07,
|
| 1222 |
+
"logits/chosen": -1.5152695178985596,
|
| 1223 |
+
"logits/rejected": -1.423516869544983,
|
| 1224 |
+
"logps/chosen": -439.678955078125,
|
| 1225 |
+
"logps/rejected": -520.2861938476562,
|
| 1226 |
+
"loss": 0.5025,
|
| 1227 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1228 |
+
"rewards/chosen": -2.0016751289367676,
|
| 1229 |
+
"rewards/margins": 0.9704031944274902,
|
| 1230 |
+
"rewards/rejected": -2.972078561782837,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 0.7746663177178749,
|
| 1235 |
+
"grad_norm": 6.294313395553081,
|
| 1236 |
+
"learning_rate": 7.338500848029603e-07,
|
| 1237 |
+
"logits/chosen": -1.503118872642517,
|
| 1238 |
+
"logits/rejected": -1.4224625825881958,
|
| 1239 |
+
"logps/chosen": -494.43505859375,
|
| 1240 |
+
"logps/rejected": -529.97607421875,
|
| 1241 |
+
"loss": 0.5112,
|
| 1242 |
+
"rewards/accuracies": 0.75,
|
| 1243 |
+
"rewards/chosen": -2.1212410926818848,
|
| 1244 |
+
"rewards/margins": 0.9084514379501343,
|
| 1245 |
+
"rewards/rejected": -3.0296926498413086,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 0.7851347814708192,
|
| 1250 |
+
"grad_norm": 4.456878260756767,
|
| 1251 |
+
"learning_rate": 6.70334495204884e-07,
|
| 1252 |
+
"logits/chosen": -1.3774914741516113,
|
| 1253 |
+
"logits/rejected": -1.3466551303863525,
|
| 1254 |
+
"logps/chosen": -453.56060791015625,
|
| 1255 |
+
"logps/rejected": -562.9486083984375,
|
| 1256 |
+
"loss": 0.4843,
|
| 1257 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1258 |
+
"rewards/chosen": -2.2453207969665527,
|
| 1259 |
+
"rewards/margins": 1.0070064067840576,
|
| 1260 |
+
"rewards/rejected": -3.2523269653320312,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.7956032452237635,
|
| 1265 |
+
"grad_norm": 4.6479649573926425,
|
| 1266 |
+
"learning_rate": 6.092659210462232e-07,
|
| 1267 |
+
"logits/chosen": -1.4880906343460083,
|
| 1268 |
+
"logits/rejected": -1.3924858570098877,
|
| 1269 |
+
"logps/chosen": -477.60284423828125,
|
| 1270 |
+
"logps/rejected": -531.3121337890625,
|
| 1271 |
+
"loss": 0.508,
|
| 1272 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1273 |
+
"rewards/chosen": -2.1819820404052734,
|
| 1274 |
+
"rewards/margins": 0.885240912437439,
|
| 1275 |
+
"rewards/rejected": -3.067223072052002,
|
| 1276 |
+
"step": 760
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.8060717089767077,
|
| 1280 |
+
"grad_norm": 3.847763381152239,
|
| 1281 |
+
"learning_rate": 5.507260361320738e-07,
|
| 1282 |
+
"logits/chosen": -1.5711116790771484,
|
| 1283 |
+
"logits/rejected": -1.5567753314971924,
|
| 1284 |
+
"logps/chosen": -490.78106689453125,
|
| 1285 |
+
"logps/rejected": -591.5490112304688,
|
| 1286 |
+
"loss": 0.4728,
|
| 1287 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1288 |
+
"rewards/chosen": -2.039038896560669,
|
| 1289 |
+
"rewards/margins": 0.9788064956665039,
|
| 1290 |
+
"rewards/rejected": -3.017845630645752,
|
| 1291 |
+
"step": 770
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.816540172729652,
|
| 1295 |
+
"grad_norm": 6.007803688457658,
|
| 1296 |
+
"learning_rate": 4.947931323697983e-07,
|
| 1297 |
+
"logits/chosen": -1.5654757022857666,
|
| 1298 |
+
"logits/rejected": -1.4224697351455688,
|
| 1299 |
+
"logps/chosen": -507.5931091308594,
|
| 1300 |
+
"logps/rejected": -524.8499145507812,
|
| 1301 |
+
"loss": 0.4961,
|
| 1302 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 1303 |
+
"rewards/chosen": -1.997415542602539,
|
| 1304 |
+
"rewards/margins": 0.7785658240318298,
|
| 1305 |
+
"rewards/rejected": -2.7759814262390137,
|
| 1306 |
+
"step": 780
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.8270086364825961,
|
| 1310 |
+
"grad_norm": 5.439843899605622,
|
| 1311 |
+
"learning_rate": 4.4154201506053985e-07,
|
| 1312 |
+
"logits/chosen": -1.4940153360366821,
|
| 1313 |
+
"logits/rejected": -1.4301998615264893,
|
| 1314 |
+
"logps/chosen": -463.53424072265625,
|
| 1315 |
+
"logps/rejected": -545.1070556640625,
|
| 1316 |
+
"loss": 0.5221,
|
| 1317 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1318 |
+
"rewards/chosen": -2.205313205718994,
|
| 1319 |
+
"rewards/margins": 0.8524004817008972,
|
| 1320 |
+
"rewards/rejected": -3.057713747024536,
|
| 1321 |
+
"step": 790
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.8374771002355405,
|
| 1325 |
+
"grad_norm": 6.835477486779152,
|
| 1326 |
+
"learning_rate": 3.910439028537638e-07,
|
| 1327 |
+
"logits/chosen": -1.4495314359664917,
|
| 1328 |
+
"logits/rejected": -1.433584451675415,
|
| 1329 |
+
"logps/chosen": -439.5552673339844,
|
| 1330 |
+
"logps/rejected": -554.7761840820312,
|
| 1331 |
+
"loss": 0.4898,
|
| 1332 |
+
"rewards/accuracies": 0.793749988079071,
|
| 1333 |
+
"rewards/chosen": -2.071547031402588,
|
| 1334 |
+
"rewards/margins": 1.1167417764663696,
|
| 1335 |
+
"rewards/rejected": -3.188288927078247,
|
| 1336 |
+
"step": 800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 0.8374771002355405,
|
| 1340 |
+
"eval_logits/chosen": -1.4946216344833374,
|
| 1341 |
+
"eval_logits/rejected": -1.427457332611084,
|
| 1342 |
+
"eval_logps/chosen": -470.2820739746094,
|
| 1343 |
+
"eval_logps/rejected": -547.9973754882812,
|
| 1344 |
+
"eval_loss": 0.4948446750640869,
|
| 1345 |
+
"eval_rewards/accuracies": 0.7698412537574768,
|
| 1346 |
+
"eval_rewards/chosen": -2.0400798320770264,
|
| 1347 |
+
"eval_rewards/margins": 0.9715465903282166,
|
| 1348 |
+
"eval_rewards/rejected": -3.0116262435913086,
|
| 1349 |
+
"eval_runtime": 277.1008,
|
| 1350 |
+
"eval_samples_per_second": 7.218,
|
| 1351 |
+
"eval_steps_per_second": 0.227,
|
| 1352 |
+
"step": 800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.8479455639884846,
|
| 1356 |
+
"grad_norm": 4.180571004247573,
|
| 1357 |
+
"learning_rate": 3.4336633249862084e-07,
|
| 1358 |
+
"logits/chosen": -1.5085508823394775,
|
| 1359 |
+
"logits/rejected": -1.4109275341033936,
|
| 1360 |
+
"logps/chosen": -499.6455993652344,
|
| 1361 |
+
"logps/rejected": -552.9725341796875,
|
| 1362 |
+
"loss": 0.5021,
|
| 1363 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1364 |
+
"rewards/chosen": -2.0836286544799805,
|
| 1365 |
+
"rewards/margins": 0.9168750643730164,
|
| 1366 |
+
"rewards/rejected": -3.0005037784576416,
|
| 1367 |
+
"step": 810
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.8584140277414289,
|
| 1371 |
+
"grad_norm": 5.175742678334051,
|
| 1372 |
+
"learning_rate": 2.98573068519539e-07,
|
| 1373 |
+
"logits/chosen": -1.548295021057129,
|
| 1374 |
+
"logits/rejected": -1.5127002000808716,
|
| 1375 |
+
"logps/chosen": -480.94427490234375,
|
| 1376 |
+
"logps/rejected": -558.2870483398438,
|
| 1377 |
+
"loss": 0.5078,
|
| 1378 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1379 |
+
"rewards/chosen": -2.1707310676574707,
|
| 1380 |
+
"rewards/margins": 0.8862228393554688,
|
| 1381 |
+
"rewards/rejected": -3.0569539070129395,
|
| 1382 |
+
"step": 820
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.8688824914943732,
|
| 1386 |
+
"grad_norm": 5.516648759407251,
|
| 1387 |
+
"learning_rate": 2.5672401793681854e-07,
|
| 1388 |
+
"logits/chosen": -1.4796596765518188,
|
| 1389 |
+
"logits/rejected": -1.461284875869751,
|
| 1390 |
+
"logps/chosen": -461.5376892089844,
|
| 1391 |
+
"logps/rejected": -552.8976440429688,
|
| 1392 |
+
"loss": 0.5097,
|
| 1393 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1394 |
+
"rewards/chosen": -2.1345925331115723,
|
| 1395 |
+
"rewards/margins": 0.9002341032028198,
|
| 1396 |
+
"rewards/rejected": -3.0348267555236816,
|
| 1397 |
+
"step": 830
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.8793509552473174,
|
| 1401 |
+
"grad_norm": 4.741723004470474,
|
| 1402 |
+
"learning_rate": 2.178751501463036e-07,
|
| 1403 |
+
"logits/chosen": -1.502801537513733,
|
| 1404 |
+
"logits/rejected": -1.4475767612457275,
|
| 1405 |
+
"logps/chosen": -501.8017578125,
|
| 1406 |
+
"logps/rejected": -600.8604125976562,
|
| 1407 |
+
"loss": 0.4766,
|
| 1408 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1409 |
+
"rewards/chosen": -2.2499661445617676,
|
| 1410 |
+
"rewards/margins": 1.028209924697876,
|
| 1411 |
+
"rewards/rejected": -3.2781760692596436,
|
| 1412 |
+
"step": 840
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.8898194190002617,
|
| 1416 |
+
"grad_norm": 5.522067224662803,
|
| 1417 |
+
"learning_rate": 1.820784220652766e-07,
|
| 1418 |
+
"logits/chosen": -1.454413652420044,
|
| 1419 |
+
"logits/rejected": -1.3597214221954346,
|
| 1420 |
+
"logps/chosen": -481.33905029296875,
|
| 1421 |
+
"logps/rejected": -549.5355224609375,
|
| 1422 |
+
"loss": 0.4602,
|
| 1423 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1424 |
+
"rewards/chosen": -2.1423401832580566,
|
| 1425 |
+
"rewards/margins": 0.996941864490509,
|
| 1426 |
+
"rewards/rejected": -3.139281988143921,
|
| 1427 |
+
"step": 850
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.9002878827532059,
|
| 1431 |
+
"grad_norm": 5.0423547871768575,
|
| 1432 |
+
"learning_rate": 1.4938170864468636e-07,
|
| 1433 |
+
"logits/chosen": -1.5176475048065186,
|
| 1434 |
+
"logits/rejected": -1.414750337600708,
|
| 1435 |
+
"logps/chosen": -488.75177001953125,
|
| 1436 |
+
"logps/rejected": -550.3502197265625,
|
| 1437 |
+
"loss": 0.4715,
|
| 1438 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1439 |
+
"rewards/chosen": -2.1736044883728027,
|
| 1440 |
+
"rewards/margins": 0.8499178886413574,
|
| 1441 |
+
"rewards/rejected": -3.0235226154327393,
|
| 1442 |
+
"step": 860
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.9107563465061502,
|
| 1446 |
+
"grad_norm": 4.698698474333166,
|
| 1447 |
+
"learning_rate": 1.1982873884064466e-07,
|
| 1448 |
+
"logits/chosen": -1.5513103008270264,
|
| 1449 |
+
"logits/rejected": -1.4094650745391846,
|
| 1450 |
+
"logps/chosen": -506.3067321777344,
|
| 1451 |
+
"logps/rejected": -559.7911376953125,
|
| 1452 |
+
"loss": 0.4896,
|
| 1453 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1454 |
+
"rewards/chosen": -2.2394676208496094,
|
| 1455 |
+
"rewards/margins": 1.060282588005066,
|
| 1456 |
+
"rewards/rejected": -3.2997500896453857,
|
| 1457 |
+
"step": 870
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.9212248102590945,
|
| 1461 |
+
"grad_norm": 4.851477312614225,
|
| 1462 |
+
"learning_rate": 9.345903713082305e-08,
|
| 1463 |
+
"logits/chosen": -1.5655453205108643,
|
| 1464 |
+
"logits/rejected": -1.4765007495880127,
|
| 1465 |
+
"logps/chosen": -480.64605712890625,
|
| 1466 |
+
"logps/rejected": -537.4286499023438,
|
| 1467 |
+
"loss": 0.4762,
|
| 1468 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1469 |
+
"rewards/chosen": -2.0842466354370117,
|
| 1470 |
+
"rewards/margins": 0.9646340608596802,
|
| 1471 |
+
"rewards/rejected": -3.0488810539245605,
|
| 1472 |
+
"step": 880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.9316932740120387,
|
| 1476 |
+
"grad_norm": 5.226575451994075,
|
| 1477 |
+
"learning_rate": 7.030787065396866e-08,
|
| 1478 |
+
"logits/chosen": -1.4315617084503174,
|
| 1479 |
+
"logits/rejected": -1.373170256614685,
|
| 1480 |
+
"logps/chosen": -466.0337829589844,
|
| 1481 |
+
"logps/rejected": -583.4358520507812,
|
| 1482 |
+
"loss": 0.4857,
|
| 1483 |
+
"rewards/accuracies": 0.793749988079071,
|
| 1484 |
+
"rewards/chosen": -2.1631832122802734,
|
| 1485 |
+
"rewards/margins": 1.2467734813690186,
|
| 1486 |
+
"rewards/rejected": -3.409956455230713,
|
| 1487 |
+
"step": 890
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.942161737764983,
|
| 1491 |
+
"grad_norm": 5.514107153250047,
|
| 1492 |
+
"learning_rate": 5.0406202043228604e-08,
|
| 1493 |
+
"logits/chosen": -1.3586971759796143,
|
| 1494 |
+
"logits/rejected": -1.3444792032241821,
|
| 1495 |
+
"logps/chosen": -478.5252380371094,
|
| 1496 |
+
"logps/rejected": -628.2623291015625,
|
| 1497 |
+
"loss": 0.4785,
|
| 1498 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1499 |
+
"rewards/chosen": -2.112252712249756,
|
| 1500 |
+
"rewards/margins": 1.2148876190185547,
|
| 1501 |
+
"rewards/rejected": -3.3271403312683105,
|
| 1502 |
+
"step": 900
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.942161737764983,
|
| 1506 |
+
"eval_logits/chosen": -1.4777917861938477,
|
| 1507 |
+
"eval_logits/rejected": -1.4105483293533325,
|
| 1508 |
+
"eval_logps/chosen": -483.4023742675781,
|
| 1509 |
+
"eval_logps/rejected": -564.8470458984375,
|
| 1510 |
+
"eval_loss": 0.493284672498703,
|
| 1511 |
+
"eval_rewards/accuracies": 0.773809552192688,
|
| 1512 |
+
"eval_rewards/chosen": -2.1712825298309326,
|
| 1513 |
+
"eval_rewards/margins": 1.0088402032852173,
|
| 1514 |
+
"eval_rewards/rejected": -3.1801228523254395,
|
| 1515 |
+
"eval_runtime": 276.8996,
|
| 1516 |
+
"eval_samples_per_second": 7.223,
|
| 1517 |
+
"eval_steps_per_second": 0.228,
|
| 1518 |
+
"step": 900
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 0.9526302015179272,
|
| 1522 |
+
"grad_norm": 7.959876344290962,
|
| 1523 |
+
"learning_rate": 3.378064801637687e-08,
|
| 1524 |
+
"logits/chosen": -1.476231336593628,
|
| 1525 |
+
"logits/rejected": -1.382854700088501,
|
| 1526 |
+
"logps/chosen": -463.75653076171875,
|
| 1527 |
+
"logps/rejected": -513.1383666992188,
|
| 1528 |
+
"loss": 0.5014,
|
| 1529 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1530 |
+
"rewards/chosen": -2.1452784538269043,
|
| 1531 |
+
"rewards/margins": 0.9462319612503052,
|
| 1532 |
+
"rewards/rejected": -3.09151029586792,
|
| 1533 |
+
"step": 910
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.9630986652708715,
|
| 1537 |
+
"grad_norm": 10.786910308077847,
|
| 1538 |
+
"learning_rate": 2.0453443778310766e-08,
|
| 1539 |
+
"logits/chosen": -1.4654277563095093,
|
| 1540 |
+
"logits/rejected": -1.371619462966919,
|
| 1541 |
+
"logps/chosen": -502.68450927734375,
|
| 1542 |
+
"logps/rejected": -547.508544921875,
|
| 1543 |
+
"loss": 0.5041,
|
| 1544 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1545 |
+
"rewards/chosen": -2.2171082496643066,
|
| 1546 |
+
"rewards/margins": 0.9030927419662476,
|
| 1547 |
+
"rewards/rejected": -3.1202011108398438,
|
| 1548 |
+
"step": 920
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 0.9735671290238157,
|
| 1552 |
+
"grad_norm": 5.431830769761952,
|
| 1553 |
+
"learning_rate": 1.0442413283435759e-08,
|
| 1554 |
+
"logits/chosen": -1.4298330545425415,
|
| 1555 |
+
"logits/rejected": -1.3073384761810303,
|
| 1556 |
+
"logps/chosen": -518.3793334960938,
|
| 1557 |
+
"logps/rejected": -556.9884033203125,
|
| 1558 |
+
"loss": 0.4886,
|
| 1559 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1560 |
+
"rewards/chosen": -2.1807284355163574,
|
| 1561 |
+
"rewards/margins": 0.9955088496208191,
|
| 1562 |
+
"rewards/rejected": -3.176237106323242,
|
| 1563 |
+
"step": 930
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.98403559277676,
|
| 1567 |
+
"grad_norm": 5.391800635976396,
|
| 1568 |
+
"learning_rate": 3.760945397705828e-09,
|
| 1569 |
+
"logits/chosen": -1.457722544670105,
|
| 1570 |
+
"logits/rejected": -1.4312618970870972,
|
| 1571 |
+
"logps/chosen": -479.5771484375,
|
| 1572 |
+
"logps/rejected": -562.9434814453125,
|
| 1573 |
+
"loss": 0.4851,
|
| 1574 |
+
"rewards/accuracies": 0.699999988079071,
|
| 1575 |
+
"rewards/chosen": -2.1574504375457764,
|
| 1576 |
+
"rewards/margins": 0.8255201578140259,
|
| 1577 |
+
"rewards/rejected": -2.982970714569092,
|
| 1578 |
+
"step": 940
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.9945040565297043,
|
| 1582 |
+
"grad_norm": 5.016778527628419,
|
| 1583 |
+
"learning_rate": 4.1797599220405605e-10,
|
| 1584 |
+
"logits/chosen": -1.4823999404907227,
|
| 1585 |
+
"logits/rejected": -1.3737461566925049,
|
| 1586 |
+
"logps/chosen": -480.68951416015625,
|
| 1587 |
+
"logps/rejected": -539.4514770507812,
|
| 1588 |
+
"loss": 0.4874,
|
| 1589 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1590 |
+
"rewards/chosen": -2.134705066680908,
|
| 1591 |
+
"rewards/margins": 0.8970800638198853,
|
| 1592 |
+
"rewards/rejected": -3.031785249710083,
|
| 1593 |
+
"step": 950
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.9997382884061764,
|
| 1597 |
+
"step": 955,
|
| 1598 |
+
"total_flos": 0.0,
|
| 1599 |
+
"train_loss": 0.5333428270529702,
|
| 1600 |
+
"train_runtime": 30179.1373,
|
| 1601 |
+
"train_samples_per_second": 2.026,
|
| 1602 |
+
"train_steps_per_second": 0.032
|
| 1603 |
+
}
|
| 1604 |
+
],
|
| 1605 |
+
"logging_steps": 10,
|
| 1606 |
+
"max_steps": 955,
|
| 1607 |
+
"num_input_tokens_seen": 0,
|
| 1608 |
+
"num_train_epochs": 1,
|
| 1609 |
+
"save_steps": 100,
|
| 1610 |
+
"stateful_callbacks": {
|
| 1611 |
+
"TrainerControl": {
|
| 1612 |
+
"args": {
|
| 1613 |
+
"should_epoch_stop": false,
|
| 1614 |
+
"should_evaluate": false,
|
| 1615 |
+
"should_log": false,
|
| 1616 |
+
"should_save": true,
|
| 1617 |
+
"should_training_stop": false
|
| 1618 |
+
},
|
| 1619 |
+
"attributes": {}
|
| 1620 |
+
}
|
| 1621 |
+
},
|
| 1622 |
+
"total_flos": 0.0,
|
| 1623 |
+
"train_batch_size": 4,
|
| 1624 |
+
"trial_name": null,
|
| 1625 |
+
"trial_params": null
|
| 1626 |
+
}
|