Commit
•
844f260
1
Parent(s):
e303bc6
Model save
Browse files- README.md +83 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1852 -0
README.md
ADDED
|
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: slm-research-vn/Qwen2-7B-Instruct-SPPO-Function-call-v2.5
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: Qwen2-7B-Instruct-SPPO-Function-call-v2.6
|
| 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 |
+
# Qwen2-7B-Instruct-SPPO-Function-call-v2.6
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [slm-research-vn/Qwen2-7B-Instruct-SPPO-Function-call-v2.5](https://huggingface.co/slm-research-vn/Qwen2-7B-Instruct-SPPO-Function-call-v2.5) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.2999
|
| 21 |
+
- Rewards/chosen: 1.6798
|
| 22 |
+
- Rewards/rejected: -0.4929
|
| 23 |
+
- Rewards/accuracies: 0.8844
|
| 24 |
+
- Rewards/margins: 2.1726
|
| 25 |
+
- Logps/rejected: -276.8312
|
| 26 |
+
- Logps/chosen: -200.8157
|
| 27 |
+
- Logits/rejected: -0.6690
|
| 28 |
+
- Logits/chosen: -0.6635
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 1e-06
|
| 48 |
+
- train_batch_size: 1
|
| 49 |
+
- eval_batch_size: 1
|
| 50 |
+
- seed: 42
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 8
|
| 53 |
+
- gradient_accumulation_steps: 4
|
| 54 |
+
- total_train_batch_size: 32
|
| 55 |
+
- total_eval_batch_size: 8
|
| 56 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
+
- lr_scheduler_type: cosine
|
| 58 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 59 |
+
- num_epochs: 1
|
| 60 |
+
|
| 61 |
+
### Training results
|
| 62 |
+
|
| 63 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 64 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 65 |
+
| 0.6437 | 0.0916 | 100 | 0.6128 | 0.3050 | 0.0739 | 0.7254 | 0.2311 | -265.4963 | -228.3116 | -0.7319 | -0.7206 |
|
| 66 |
+
| 0.5175 | 0.1832 | 200 | 0.4987 | 1.1265 | 0.2914 | 0.8237 | 0.8351 | -261.1460 | -211.8815 | -0.7134 | -0.7068 |
|
| 67 |
+
| 0.3903 | 0.2749 | 300 | 0.4279 | 1.7297 | 0.4889 | 0.8468 | 1.2408 | -257.1960 | -199.8173 | -0.6700 | -0.6642 |
|
| 68 |
+
| 0.3712 | 0.3665 | 400 | 0.3781 | 1.7272 | 0.2255 | 0.8468 | 1.5017 | -262.4645 | -199.8672 | -0.6756 | -0.6691 |
|
| 69 |
+
| 0.3064 | 0.4581 | 500 | 0.3477 | 1.7220 | -0.0183 | 0.8613 | 1.7403 | -267.3389 | -199.9704 | -0.6642 | -0.6488 |
|
| 70 |
+
| 0.3054 | 0.5497 | 600 | 0.3271 | 1.6469 | -0.1977 | 0.8671 | 1.8447 | -270.9281 | -201.4723 | -0.6576 | -0.6407 |
|
| 71 |
+
| 0.2919 | 0.6413 | 700 | 0.3144 | 1.7376 | -0.3034 | 0.8642 | 2.0410 | -273.0414 | -199.6590 | -0.6753 | -0.6672 |
|
| 72 |
+
| 0.314 | 0.7329 | 800 | 0.3056 | 1.7037 | -0.4229 | 0.8671 | 2.1266 | -275.4323 | -200.3379 | -0.6685 | -0.6574 |
|
| 73 |
+
| 0.3014 | 0.8246 | 900 | 0.3020 | 1.6807 | -0.4632 | 0.8699 | 2.1439 | -276.2374 | -200.7971 | -0.6702 | -0.6641 |
|
| 74 |
+
| 0.268 | 0.9162 | 1000 | 0.2999 | 1.6798 | -0.4929 | 0.8844 | 2.1726 | -276.8312 | -200.8157 | -0.6690 | -0.6635 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- PEFT 0.12.0
|
| 80 |
+
- Transformers 4.44.0
|
| 81 |
+
- Pytorch 2.3.1+cu121
|
| 82 |
+
- Datasets 2.20.0
|
| 83 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9995419147961521,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.37405444834616947,
|
| 5 |
+
"train_runtime": 8738.9435,
|
| 6 |
+
"train_samples": 34924,
|
| 7 |
+
"train_samples_per_second": 3.996,
|
| 8 |
+
"train_steps_per_second": 0.125
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9995419147961521,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.37405444834616947,
|
| 5 |
+
"train_runtime": 8738.9435,
|
| 6 |
+
"train_samples": 34924,
|
| 7 |
+
"train_samples_per_second": 3.996,
|
| 8 |
+
"train_steps_per_second": 0.125
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1852 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9995419147961521,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1091,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0009161704076958314,
|
| 13 |
+
"grad_norm": 2.538351535797119,
|
| 14 |
+
"learning_rate": 9.09090909090909e-09,
|
| 15 |
+
"logits/chosen": -0.612647533416748,
|
| 16 |
+
"logits/rejected": -0.43005383014678955,
|
| 17 |
+
"logps/chosen": -269.1338195800781,
|
| 18 |
+
"logps/rejected": -265.996826171875,
|
| 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": 0.009161704076958314,
|
| 28 |
+
"grad_norm": 3.99981427192688,
|
| 29 |
+
"learning_rate": 9.09090909090909e-08,
|
| 30 |
+
"logits/chosen": -0.7947888374328613,
|
| 31 |
+
"logits/rejected": -0.8272799849510193,
|
| 32 |
+
"logps/chosen": -172.9255828857422,
|
| 33 |
+
"logps/rejected": -192.15106201171875,
|
| 34 |
+
"loss": 0.6942,
|
| 35 |
+
"rewards/accuracies": 0.5,
|
| 36 |
+
"rewards/chosen": -0.012904312461614609,
|
| 37 |
+
"rewards/margins": -0.015133652836084366,
|
| 38 |
+
"rewards/rejected": 0.002229340374469757,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.01832340815391663,
|
| 43 |
+
"grad_norm": 1.3273310661315918,
|
| 44 |
+
"learning_rate": 1.818181818181818e-07,
|
| 45 |
+
"logits/chosen": -0.7671724557876587,
|
| 46 |
+
"logits/rejected": -0.7643235921859741,
|
| 47 |
+
"logps/chosen": -177.4793243408203,
|
| 48 |
+
"logps/rejected": -201.07656860351562,
|
| 49 |
+
"loss": 0.6933,
|
| 50 |
+
"rewards/accuracies": 0.5,
|
| 51 |
+
"rewards/chosen": -0.006979703903198242,
|
| 52 |
+
"rewards/margins": -0.00390571728348732,
|
| 53 |
+
"rewards/rejected": -0.003073985455557704,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.027485112230874943,
|
| 58 |
+
"grad_norm": 2.06701922416687,
|
| 59 |
+
"learning_rate": 2.727272727272727e-07,
|
| 60 |
+
"logits/chosen": -0.7477768659591675,
|
| 61 |
+
"logits/rejected": -0.770847737789154,
|
| 62 |
+
"logps/chosen": -219.7743377685547,
|
| 63 |
+
"logps/rejected": -253.83798217773438,
|
| 64 |
+
"loss": 0.6933,
|
| 65 |
+
"rewards/accuracies": 0.574999988079071,
|
| 66 |
+
"rewards/chosen": -0.007790957577526569,
|
| 67 |
+
"rewards/margins": 0.0053612953051924706,
|
| 68 |
+
"rewards/rejected": -0.013152251951396465,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.03664681630783326,
|
| 73 |
+
"grad_norm": 2.176619052886963,
|
| 74 |
+
"learning_rate": 3.636363636363636e-07,
|
| 75 |
+
"logits/chosen": -0.6840143203735352,
|
| 76 |
+
"logits/rejected": -0.6969764828681946,
|
| 77 |
+
"logps/chosen": -159.89796447753906,
|
| 78 |
+
"logps/rejected": -218.1041717529297,
|
| 79 |
+
"loss": 0.6902,
|
| 80 |
+
"rewards/accuracies": 0.375,
|
| 81 |
+
"rewards/chosen": 0.0005666827782988548,
|
| 82 |
+
"rewards/margins": 0.012707608751952648,
|
| 83 |
+
"rewards/rejected": -0.012140927836298943,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.04580852038479157,
|
| 88 |
+
"grad_norm": 2.182387590408325,
|
| 89 |
+
"learning_rate": 4.545454545454545e-07,
|
| 90 |
+
"logits/chosen": -0.5698288083076477,
|
| 91 |
+
"logits/rejected": -0.7136921286582947,
|
| 92 |
+
"logps/chosen": -203.5851287841797,
|
| 93 |
+
"logps/rejected": -231.3826446533203,
|
| 94 |
+
"loss": 0.6878,
|
| 95 |
+
"rewards/accuracies": 0.550000011920929,
|
| 96 |
+
"rewards/chosen": 0.01093050092458725,
|
| 97 |
+
"rewards/margins": 0.021884005516767502,
|
| 98 |
+
"rewards/rejected": -0.010953502729535103,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.054970224461749886,
|
| 103 |
+
"grad_norm": 1.757957935333252,
|
| 104 |
+
"learning_rate": 5.454545454545454e-07,
|
| 105 |
+
"logits/chosen": -0.7015949487686157,
|
| 106 |
+
"logits/rejected": -0.7727667689323425,
|
| 107 |
+
"logps/chosen": -233.5804901123047,
|
| 108 |
+
"logps/rejected": -278.2530212402344,
|
| 109 |
+
"loss": 0.6876,
|
| 110 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 111 |
+
"rewards/chosen": 0.021485041826963425,
|
| 112 |
+
"rewards/margins": 0.0006370929768308997,
|
| 113 |
+
"rewards/rejected": 0.020847950130701065,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.0641319285387082,
|
| 118 |
+
"grad_norm": 2.0394017696380615,
|
| 119 |
+
"learning_rate": 6.363636363636363e-07,
|
| 120 |
+
"logits/chosen": -0.8689748644828796,
|
| 121 |
+
"logits/rejected": -0.808820366859436,
|
| 122 |
+
"logps/chosen": -147.0337371826172,
|
| 123 |
+
"logps/rejected": -209.3056182861328,
|
| 124 |
+
"loss": 0.6815,
|
| 125 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 126 |
+
"rewards/chosen": 0.05839823558926582,
|
| 127 |
+
"rewards/margins": 0.03576263412833214,
|
| 128 |
+
"rewards/rejected": 0.022635603323578835,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.07329363261566652,
|
| 133 |
+
"grad_norm": 3.2475733757019043,
|
| 134 |
+
"learning_rate": 7.272727272727272e-07,
|
| 135 |
+
"logits/chosen": -0.7972738742828369,
|
| 136 |
+
"logits/rejected": -0.8733490109443665,
|
| 137 |
+
"logps/chosen": -202.31976318359375,
|
| 138 |
+
"logps/rejected": -230.676513671875,
|
| 139 |
+
"loss": 0.6706,
|
| 140 |
+
"rewards/accuracies": 0.625,
|
| 141 |
+
"rewards/chosen": 0.0555371418595314,
|
| 142 |
+
"rewards/margins": 0.04019620642066002,
|
| 143 |
+
"rewards/rejected": 0.015340929850935936,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.08245533669262482,
|
| 148 |
+
"grad_norm": 2.1004042625427246,
|
| 149 |
+
"learning_rate": 8.181818181818182e-07,
|
| 150 |
+
"logits/chosen": -0.7756220698356628,
|
| 151 |
+
"logits/rejected": -0.7883769869804382,
|
| 152 |
+
"logps/chosen": -181.07911682128906,
|
| 153 |
+
"logps/rejected": -253.35226440429688,
|
| 154 |
+
"loss": 0.657,
|
| 155 |
+
"rewards/accuracies": 0.574999988079071,
|
| 156 |
+
"rewards/chosen": 0.09145340323448181,
|
| 157 |
+
"rewards/margins": 0.05497417598962784,
|
| 158 |
+
"rewards/rejected": 0.03647923097014427,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.09161704076958314,
|
| 163 |
+
"grad_norm": 1.711387276649475,
|
| 164 |
+
"learning_rate": 9.09090909090909e-07,
|
| 165 |
+
"logits/chosen": -0.6988880634307861,
|
| 166 |
+
"logits/rejected": -0.695970356464386,
|
| 167 |
+
"logps/chosen": -210.9344940185547,
|
| 168 |
+
"logps/rejected": -292.63885498046875,
|
| 169 |
+
"loss": 0.6437,
|
| 170 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 171 |
+
"rewards/chosen": 0.2389262169599533,
|
| 172 |
+
"rewards/margins": 0.13863424956798553,
|
| 173 |
+
"rewards/rejected": 0.10029196739196777,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.09161704076958314,
|
| 178 |
+
"eval_logits/chosen": -0.7206078171730042,
|
| 179 |
+
"eval_logits/rejected": -0.7319283485412598,
|
| 180 |
+
"eval_logps/chosen": -228.31155395507812,
|
| 181 |
+
"eval_logps/rejected": -265.4963073730469,
|
| 182 |
+
"eval_loss": 0.6128209829330444,
|
| 183 |
+
"eval_rewards/accuracies": 0.7254335284233093,
|
| 184 |
+
"eval_rewards/chosen": 0.30496734380722046,
|
| 185 |
+
"eval_rewards/margins": 0.23110172152519226,
|
| 186 |
+
"eval_rewards/rejected": 0.0738656222820282,
|
| 187 |
+
"eval_runtime": 264.1865,
|
| 188 |
+
"eval_samples_per_second": 10.459,
|
| 189 |
+
"eval_steps_per_second": 1.31,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.10077874484654145,
|
| 194 |
+
"grad_norm": 2.3657712936401367,
|
| 195 |
+
"learning_rate": 1e-06,
|
| 196 |
+
"logits/chosen": -0.9458168745040894,
|
| 197 |
+
"logits/rejected": -0.912136435508728,
|
| 198 |
+
"logps/chosen": -266.46087646484375,
|
| 199 |
+
"logps/rejected": -255.1190643310547,
|
| 200 |
+
"loss": 0.6226,
|
| 201 |
+
"rewards/accuracies": 0.675000011920929,
|
| 202 |
+
"rewards/chosen": 0.24092476069927216,
|
| 203 |
+
"rewards/margins": 0.15844842791557312,
|
| 204 |
+
"rewards/rejected": 0.08247633278369904,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.10994044892349977,
|
| 209 |
+
"grad_norm": 1.905785083770752,
|
| 210 |
+
"learning_rate": 9.997436315234263e-07,
|
| 211 |
+
"logits/chosen": -0.7388381958007812,
|
| 212 |
+
"logits/rejected": -0.7873945236206055,
|
| 213 |
+
"logps/chosen": -167.81883239746094,
|
| 214 |
+
"logps/rejected": -176.68295288085938,
|
| 215 |
+
"loss": 0.6136,
|
| 216 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 217 |
+
"rewards/chosen": 0.3122307062149048,
|
| 218 |
+
"rewards/margins": 0.20933406054973602,
|
| 219 |
+
"rewards/rejected": 0.10289661586284637,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.11910215300045808,
|
| 224 |
+
"grad_norm": 1.5484569072723389,
|
| 225 |
+
"learning_rate": 9.989747889928883e-07,
|
| 226 |
+
"logits/chosen": -0.7524275183677673,
|
| 227 |
+
"logits/rejected": -0.8292320370674133,
|
| 228 |
+
"logps/chosen": -197.0650177001953,
|
| 229 |
+
"logps/rejected": -237.57839965820312,
|
| 230 |
+
"loss": 0.5938,
|
| 231 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 232 |
+
"rewards/chosen": 0.5554567575454712,
|
| 233 |
+
"rewards/margins": 0.4092523455619812,
|
| 234 |
+
"rewards/rejected": 0.14620442688465118,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.1282638570774164,
|
| 239 |
+
"grad_norm": 1.4368153810501099,
|
| 240 |
+
"learning_rate": 9.976942608363393e-07,
|
| 241 |
+
"logits/chosen": -0.632690966129303,
|
| 242 |
+
"logits/rejected": -0.7571207284927368,
|
| 243 |
+
"logps/chosen": -173.34666442871094,
|
| 244 |
+
"logps/rejected": -211.22317504882812,
|
| 245 |
+
"loss": 0.5709,
|
| 246 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 247 |
+
"rewards/chosen": 0.58498615026474,
|
| 248 |
+
"rewards/margins": 0.40937310457229614,
|
| 249 |
+
"rewards/rejected": 0.17561307549476624,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.1374255611543747,
|
| 254 |
+
"grad_norm": 1.5536640882492065,
|
| 255 |
+
"learning_rate": 9.9590336020199e-07,
|
| 256 |
+
"logits/chosen": -0.6405803561210632,
|
| 257 |
+
"logits/rejected": -0.7340038418769836,
|
| 258 |
+
"logps/chosen": -182.8199462890625,
|
| 259 |
+
"logps/rejected": -243.0964813232422,
|
| 260 |
+
"loss": 0.5731,
|
| 261 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 262 |
+
"rewards/chosen": 0.6845705509185791,
|
| 263 |
+
"rewards/margins": 0.4665129780769348,
|
| 264 |
+
"rewards/rejected": 0.2180575579404831,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.14658726523133303,
|
| 269 |
+
"grad_norm": 1.1569141149520874,
|
| 270 |
+
"learning_rate": 9.936039236117095e-07,
|
| 271 |
+
"logits/chosen": -0.8644550442695618,
|
| 272 |
+
"logits/rejected": -0.8035632967948914,
|
| 273 |
+
"logps/chosen": -200.20794677734375,
|
| 274 |
+
"logps/rejected": -239.16915893554688,
|
| 275 |
+
"loss": 0.544,
|
| 276 |
+
"rewards/accuracies": 0.75,
|
| 277 |
+
"rewards/chosen": 0.826056957244873,
|
| 278 |
+
"rewards/margins": 0.5113142132759094,
|
| 279 |
+
"rewards/rejected": 0.3147428035736084,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.15574896930829135,
|
| 284 |
+
"grad_norm": 1.5735410451889038,
|
| 285 |
+
"learning_rate": 9.907983090777206e-07,
|
| 286 |
+
"logits/chosen": -0.7325465083122253,
|
| 287 |
+
"logits/rejected": -0.7175777554512024,
|
| 288 |
+
"logps/chosen": -195.422607421875,
|
| 289 |
+
"logps/rejected": -211.19418334960938,
|
| 290 |
+
"loss": 0.5275,
|
| 291 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 292 |
+
"rewards/chosen": 0.7453492879867554,
|
| 293 |
+
"rewards/margins": 0.558306872844696,
|
| 294 |
+
"rewards/rejected": 0.1870425045490265,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.16491067338524965,
|
| 299 |
+
"grad_norm": 1.4354184865951538,
|
| 300 |
+
"learning_rate": 9.874893936845187e-07,
|
| 301 |
+
"logits/chosen": -0.6600767374038696,
|
| 302 |
+
"logits/rejected": -0.7184512615203857,
|
| 303 |
+
"logps/chosen": -214.3909149169922,
|
| 304 |
+
"logps/rejected": -291.69512939453125,
|
| 305 |
+
"loss": 0.5309,
|
| 306 |
+
"rewards/accuracies": 0.925000011920929,
|
| 307 |
+
"rewards/chosen": 0.8096879124641418,
|
| 308 |
+
"rewards/margins": 0.5819981098175049,
|
| 309 |
+
"rewards/rejected": 0.2276897132396698,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.17407237746220797,
|
| 314 |
+
"grad_norm": 1.9378727674484253,
|
| 315 |
+
"learning_rate": 9.836805706384983e-07,
|
| 316 |
+
"logits/chosen": -0.796898603439331,
|
| 317 |
+
"logits/rejected": -0.7903488874435425,
|
| 318 |
+
"logps/chosen": -151.75074768066406,
|
| 319 |
+
"logps/rejected": -193.89720153808594,
|
| 320 |
+
"loss": 0.4979,
|
| 321 |
+
"rewards/accuracies": 0.824999988079071,
|
| 322 |
+
"rewards/chosen": 0.9862753748893738,
|
| 323 |
+
"rewards/margins": 0.6403971910476685,
|
| 324 |
+
"rewards/rejected": 0.3458781838417053,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.1832340815391663,
|
| 329 |
+
"grad_norm": 1.368870735168457,
|
| 330 |
+
"learning_rate": 9.793757457883061e-07,
|
| 331 |
+
"logits/chosen": -0.7233944535255432,
|
| 332 |
+
"logits/rejected": -0.7641991972923279,
|
| 333 |
+
"logps/chosen": -131.52737426757812,
|
| 334 |
+
"logps/rejected": -177.9940185546875,
|
| 335 |
+
"loss": 0.5175,
|
| 336 |
+
"rewards/accuracies": 0.875,
|
| 337 |
+
"rewards/chosen": 0.935263454914093,
|
| 338 |
+
"rewards/margins": 0.7129745483398438,
|
| 339 |
+
"rewards/rejected": 0.2222888469696045,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.1832340815391663,
|
| 344 |
+
"eval_logits/chosen": -0.7067926526069641,
|
| 345 |
+
"eval_logits/rejected": -0.7133929133415222,
|
| 346 |
+
"eval_logps/chosen": -211.8815460205078,
|
| 347 |
+
"eval_logps/rejected": -261.1459655761719,
|
| 348 |
+
"eval_loss": 0.4986709654331207,
|
| 349 |
+
"eval_rewards/accuracies": 0.823699414730072,
|
| 350 |
+
"eval_rewards/chosen": 1.1264668703079224,
|
| 351 |
+
"eval_rewards/margins": 0.8350852131843567,
|
| 352 |
+
"eval_rewards/rejected": 0.29138168692588806,
|
| 353 |
+
"eval_runtime": 253.2388,
|
| 354 |
+
"eval_samples_per_second": 10.911,
|
| 355 |
+
"eval_steps_per_second": 1.366,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.1923957856161246,
|
| 360 |
+
"grad_norm": 1.181227207183838,
|
| 361 |
+
"learning_rate": 9.745793336194975e-07,
|
| 362 |
+
"logits/chosen": -0.744937539100647,
|
| 363 |
+
"logits/rejected": -0.753220796585083,
|
| 364 |
+
"logps/chosen": -157.19473266601562,
|
| 365 |
+
"logps/rejected": -238.6236114501953,
|
| 366 |
+
"loss": 0.485,
|
| 367 |
+
"rewards/accuracies": 0.824999988079071,
|
| 368 |
+
"rewards/chosen": 0.9487007260322571,
|
| 369 |
+
"rewards/margins": 0.5822451710700989,
|
| 370 |
+
"rewards/rejected": 0.366455614566803,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.2015574896930829,
|
| 375 |
+
"grad_norm": 2.322333812713623,
|
| 376 |
+
"learning_rate": 9.69296252727595e-07,
|
| 377 |
+
"logits/chosen": -0.7686578035354614,
|
| 378 |
+
"logits/rejected": -0.7208544015884399,
|
| 379 |
+
"logps/chosen": -161.76914978027344,
|
| 380 |
+
"logps/rejected": -203.6949005126953,
|
| 381 |
+
"loss": 0.486,
|
| 382 |
+
"rewards/accuracies": 0.800000011920929,
|
| 383 |
+
"rewards/chosen": 1.1700208187103271,
|
| 384 |
+
"rewards/margins": 0.8085041046142578,
|
| 385 |
+
"rewards/rejected": 0.36151671409606934,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.21071919377004122,
|
| 390 |
+
"grad_norm": 3.2851109504699707,
|
| 391 |
+
"learning_rate": 9.63531920774199e-07,
|
| 392 |
+
"logits/chosen": -0.8623464703559875,
|
| 393 |
+
"logits/rejected": -0.864261269569397,
|
| 394 |
+
"logps/chosen": -132.91622924804688,
|
| 395 |
+
"logps/rejected": -190.5310516357422,
|
| 396 |
+
"loss": 0.4889,
|
| 397 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 398 |
+
"rewards/chosen": 1.2793647050857544,
|
| 399 |
+
"rewards/margins": 0.8540099859237671,
|
| 400 |
+
"rewards/rejected": 0.4253547787666321,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.21988089784699955,
|
| 405 |
+
"grad_norm": 2.092069387435913,
|
| 406 |
+
"learning_rate": 9.572922489313142e-07,
|
| 407 |
+
"logits/chosen": -0.8346965909004211,
|
| 408 |
+
"logits/rejected": -0.8602925539016724,
|
| 409 |
+
"logps/chosen": -177.9022216796875,
|
| 410 |
+
"logps/rejected": -209.01220703125,
|
| 411 |
+
"loss": 0.4294,
|
| 412 |
+
"rewards/accuracies": 0.75,
|
| 413 |
+
"rewards/chosen": 1.2863993644714355,
|
| 414 |
+
"rewards/margins": 0.5576584935188293,
|
| 415 |
+
"rewards/rejected": 0.7287408709526062,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.22904260192395787,
|
| 420 |
+
"grad_norm": 1.2932628393173218,
|
| 421 |
+
"learning_rate": 9.505836358195993e-07,
|
| 422 |
+
"logits/chosen": -0.7524776458740234,
|
| 423 |
+
"logits/rejected": -0.8281770944595337,
|
| 424 |
+
"logps/chosen": -144.74905395507812,
|
| 425 |
+
"logps/rejected": -226.6658172607422,
|
| 426 |
+
"loss": 0.4293,
|
| 427 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 428 |
+
"rewards/chosen": 1.3644707202911377,
|
| 429 |
+
"rewards/margins": 0.8554983139038086,
|
| 430 |
+
"rewards/rejected": 0.5089724659919739,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.23820430600091616,
|
| 435 |
+
"grad_norm": 1.3950133323669434,
|
| 436 |
+
"learning_rate": 9.434129609467483e-07,
|
| 437 |
+
"logits/chosen": -0.6876672506332397,
|
| 438 |
+
"logits/rejected": -0.6876662969589233,
|
| 439 |
+
"logps/chosen": -263.18585205078125,
|
| 440 |
+
"logps/rejected": -264.4542541503906,
|
| 441 |
+
"loss": 0.4518,
|
| 442 |
+
"rewards/accuracies": 0.824999988079071,
|
| 443 |
+
"rewards/chosen": 1.4175243377685547,
|
| 444 |
+
"rewards/margins": 0.8856536746025085,
|
| 445 |
+
"rewards/rejected": 0.5318707227706909,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.24736601007787448,
|
| 450 |
+
"grad_norm": 1.4819687604904175,
|
| 451 |
+
"learning_rate": 9.357875776527333e-07,
|
| 452 |
+
"logits/chosen": -0.6820667386054993,
|
| 453 |
+
"logits/rejected": -0.6322587132453918,
|
| 454 |
+
"logps/chosen": -173.28085327148438,
|
| 455 |
+
"logps/rejected": -196.96194458007812,
|
| 456 |
+
"loss": 0.4433,
|
| 457 |
+
"rewards/accuracies": 0.875,
|
| 458 |
+
"rewards/chosen": 1.1709938049316406,
|
| 459 |
+
"rewards/margins": 0.7361399531364441,
|
| 460 |
+
"rewards/rejected": 0.4348538815975189,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.2565277141548328,
|
| 465 |
+
"grad_norm": 1.2729604244232178,
|
| 466 |
+
"learning_rate": 9.27715305569148e-07,
|
| 467 |
+
"logits/chosen": -0.6576655507087708,
|
| 468 |
+
"logits/rejected": -0.6413074731826782,
|
| 469 |
+
"logps/chosen": -159.5121307373047,
|
| 470 |
+
"logps/rejected": -207.47195434570312,
|
| 471 |
+
"loss": 0.4029,
|
| 472 |
+
"rewards/accuracies": 0.925000011920929,
|
| 473 |
+
"rewards/chosen": 1.8569538593292236,
|
| 474 |
+
"rewards/margins": 1.4031217098236084,
|
| 475 |
+
"rewards/rejected": 0.45383185148239136,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.2656894182317911,
|
| 480 |
+
"grad_norm": 1.4309152364730835,
|
| 481 |
+
"learning_rate": 9.192044226003788e-07,
|
| 482 |
+
"logits/chosen": -0.7165778875350952,
|
| 483 |
+
"logits/rejected": -0.7207854390144348,
|
| 484 |
+
"logps/chosen": -171.72616577148438,
|
| 485 |
+
"logps/rejected": -212.6331024169922,
|
| 486 |
+
"loss": 0.4505,
|
| 487 |
+
"rewards/accuracies": 0.824999988079071,
|
| 488 |
+
"rewards/chosen": 1.8912508487701416,
|
| 489 |
+
"rewards/margins": 1.2686805725097656,
|
| 490 |
+
"rewards/rejected": 0.6225701570510864,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.2748511223087494,
|
| 495 |
+
"grad_norm": 1.5633419752120972,
|
| 496 |
+
"learning_rate": 9.102636564348294e-07,
|
| 497 |
+
"logits/chosen": -0.5978932976722717,
|
| 498 |
+
"logits/rejected": -0.7183943390846252,
|
| 499 |
+
"logps/chosen": -173.94984436035156,
|
| 500 |
+
"logps/rejected": -202.29295349121094,
|
| 501 |
+
"loss": 0.3903,
|
| 502 |
+
"rewards/accuracies": 0.949999988079071,
|
| 503 |
+
"rewards/chosen": 1.945762276649475,
|
| 504 |
+
"rewards/margins": 1.3720638751983643,
|
| 505 |
+
"rewards/rejected": 0.5736981630325317,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.2748511223087494,
|
| 510 |
+
"eval_logits/chosen": -0.6642194390296936,
|
| 511 |
+
"eval_logits/rejected": -0.6700440049171448,
|
| 512 |
+
"eval_logps/chosen": -199.8172607421875,
|
| 513 |
+
"eval_logps/rejected": -257.196044921875,
|
| 514 |
+
"eval_loss": 0.4279369115829468,
|
| 515 |
+
"eval_rewards/accuracies": 0.8468208312988281,
|
| 516 |
+
"eval_rewards/chosen": 1.7296818494796753,
|
| 517 |
+
"eval_rewards/margins": 1.240803837776184,
|
| 518 |
+
"eval_rewards/rejected": 0.48887789249420166,
|
| 519 |
+
"eval_runtime": 253.1385,
|
| 520 |
+
"eval_samples_per_second": 10.915,
|
| 521 |
+
"eval_steps_per_second": 1.367,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.28401282638570774,
|
| 526 |
+
"grad_norm": 1.3390766382217407,
|
| 527 |
+
"learning_rate": 9.009021755949051e-07,
|
| 528 |
+
"logits/chosen": -0.6982103586196899,
|
| 529 |
+
"logits/rejected": -0.7174701690673828,
|
| 530 |
+
"logps/chosen": -159.46409606933594,
|
| 531 |
+
"logps/rejected": -160.19491577148438,
|
| 532 |
+
"loss": 0.4083,
|
| 533 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 534 |
+
"rewards/chosen": 1.7644752264022827,
|
| 535 |
+
"rewards/margins": 1.0205423831939697,
|
| 536 |
+
"rewards/rejected": 0.7439330220222473,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.29317453046266606,
|
| 541 |
+
"grad_norm": 1.3620954751968384,
|
| 542 |
+
"learning_rate": 8.911295800349314e-07,
|
| 543 |
+
"logits/chosen": -0.6473032832145691,
|
| 544 |
+
"logits/rejected": -0.6676048040390015,
|
| 545 |
+
"logps/chosen": -232.72787475585938,
|
| 546 |
+
"logps/rejected": -252.5911407470703,
|
| 547 |
+
"loss": 0.4178,
|
| 548 |
+
"rewards/accuracies": 0.875,
|
| 549 |
+
"rewards/chosen": 1.7409747838974,
|
| 550 |
+
"rewards/margins": 1.134825348854065,
|
| 551 |
+
"rewards/rejected": 0.6061495542526245,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.3023362345396244,
|
| 556 |
+
"grad_norm": 0.9761648774147034,
|
| 557 |
+
"learning_rate": 8.809558912966519e-07,
|
| 558 |
+
"logits/chosen": -0.6019878387451172,
|
| 559 |
+
"logits/rejected": -0.6780513525009155,
|
| 560 |
+
"logps/chosen": -134.56356811523438,
|
| 561 |
+
"logps/rejected": -185.90939331054688,
|
| 562 |
+
"loss": 0.354,
|
| 563 |
+
"rewards/accuracies": 0.875,
|
| 564 |
+
"rewards/chosen": 2.3775978088378906,
|
| 565 |
+
"rewards/margins": 1.7786915302276611,
|
| 566 |
+
"rewards/rejected": 0.5989062786102295,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.3114979386165827,
|
| 571 |
+
"grad_norm": 1.6643030643463135,
|
| 572 |
+
"learning_rate": 8.703915422323984e-07,
|
| 573 |
+
"logits/chosen": -0.5226669907569885,
|
| 574 |
+
"logits/rejected": -0.5020606517791748,
|
| 575 |
+
"logps/chosen": -184.17132568359375,
|
| 576 |
+
"logps/rejected": -203.7013397216797,
|
| 577 |
+
"loss": 0.4022,
|
| 578 |
+
"rewards/accuracies": 0.875,
|
| 579 |
+
"rewards/chosen": 1.763214111328125,
|
| 580 |
+
"rewards/margins": 1.1016991138458252,
|
| 581 |
+
"rewards/rejected": 0.661514937877655,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.320659642693541,
|
| 586 |
+
"grad_norm": 1.045599102973938,
|
| 587 |
+
"learning_rate": 8.594473663064734e-07,
|
| 588 |
+
"logits/chosen": -0.7285621762275696,
|
| 589 |
+
"logits/rejected": -0.7740557193756104,
|
| 590 |
+
"logps/chosen": -133.10691833496094,
|
| 591 |
+
"logps/rejected": -191.98646545410156,
|
| 592 |
+
"loss": 0.3784,
|
| 593 |
+
"rewards/accuracies": 0.875,
|
| 594 |
+
"rewards/chosen": 1.8301376104354858,
|
| 595 |
+
"rewards/margins": 1.3312700986862183,
|
| 596 |
+
"rewards/rejected": 0.49886733293533325,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.3298213467704993,
|
| 601 |
+
"grad_norm": 2.110759973526001,
|
| 602 |
+
"learning_rate": 8.481345864857146e-07,
|
| 603 |
+
"logits/chosen": -0.5418592095375061,
|
| 604 |
+
"logits/rejected": -0.588280975818634,
|
| 605 |
+
"logps/chosen": -179.9706573486328,
|
| 606 |
+
"logps/rejected": -242.3267822265625,
|
| 607 |
+
"loss": 0.401,
|
| 608 |
+
"rewards/accuracies": 0.824999988079071,
|
| 609 |
+
"rewards/chosen": 1.6830288171768188,
|
| 610 |
+
"rewards/margins": 1.2477144002914429,
|
| 611 |
+
"rewards/rejected": 0.43531447649002075,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.3389830508474576,
|
| 616 |
+
"grad_norm": 1.335124135017395,
|
| 617 |
+
"learning_rate": 8.36464803730636e-07,
|
| 618 |
+
"logits/chosen": -0.8127607107162476,
|
| 619 |
+
"logits/rejected": -0.8397665023803711,
|
| 620 |
+
"logps/chosen": -143.43109130859375,
|
| 621 |
+
"logps/rejected": -185.8011932373047,
|
| 622 |
+
"loss": 0.3572,
|
| 623 |
+
"rewards/accuracies": 0.875,
|
| 624 |
+
"rewards/chosen": 1.7833560705184937,
|
| 625 |
+
"rewards/margins": 1.1723896265029907,
|
| 626 |
+
"rewards/rejected": 0.6109665632247925,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.34814475492441593,
|
| 631 |
+
"grad_norm": 1.2357772588729858,
|
| 632 |
+
"learning_rate": 8.244499850989451e-07,
|
| 633 |
+
"logits/chosen": -0.7481725811958313,
|
| 634 |
+
"logits/rejected": -0.756425678730011,
|
| 635 |
+
"logps/chosen": -117.90687561035156,
|
| 636 |
+
"logps/rejected": -198.514892578125,
|
| 637 |
+
"loss": 0.3691,
|
| 638 |
+
"rewards/accuracies": 0.875,
|
| 639 |
+
"rewards/chosen": 1.9100993871688843,
|
| 640 |
+
"rewards/margins": 1.3691816329956055,
|
| 641 |
+
"rewards/rejected": 0.5409177541732788,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.35730645900137425,
|
| 646 |
+
"grad_norm": 1.2337630987167358,
|
| 647 |
+
"learning_rate": 8.121024514736377e-07,
|
| 648 |
+
"logits/chosen": -0.5506582260131836,
|
| 649 |
+
"logits/rejected": -0.6542531251907349,
|
| 650 |
+
"logps/chosen": -116.30818176269531,
|
| 651 |
+
"logps/rejected": -186.07508850097656,
|
| 652 |
+
"loss": 0.3132,
|
| 653 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 654 |
+
"rewards/chosen": 2.1005172729492188,
|
| 655 |
+
"rewards/margins": 1.9795573949813843,
|
| 656 |
+
"rewards/rejected": 0.12095997482538223,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.3664681630783326,
|
| 661 |
+
"grad_norm": 1.261998176574707,
|
| 662 |
+
"learning_rate": 7.994348649282532e-07,
|
| 663 |
+
"logits/chosen": -0.6467943787574768,
|
| 664 |
+
"logits/rejected": -0.6284207701683044,
|
| 665 |
+
"logps/chosen": -167.55819702148438,
|
| 666 |
+
"logps/rejected": -237.6490936279297,
|
| 667 |
+
"loss": 0.3712,
|
| 668 |
+
"rewards/accuracies": 0.824999988079071,
|
| 669 |
+
"rewards/chosen": 1.9129873514175415,
|
| 670 |
+
"rewards/margins": 1.4685484170913696,
|
| 671 |
+
"rewards/rejected": 0.4444388747215271,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.3664681630783326,
|
| 676 |
+
"eval_logits/chosen": -0.6691383123397827,
|
| 677 |
+
"eval_logits/rejected": -0.6756234765052795,
|
| 678 |
+
"eval_logps/chosen": -199.86724853515625,
|
| 679 |
+
"eval_logps/rejected": -262.46453857421875,
|
| 680 |
+
"eval_loss": 0.37812188267707825,
|
| 681 |
+
"eval_rewards/accuracies": 0.8468208312988281,
|
| 682 |
+
"eval_rewards/chosen": 1.7271815538406372,
|
| 683 |
+
"eval_rewards/margins": 1.5017303228378296,
|
| 684 |
+
"eval_rewards/rejected": 0.22545117139816284,
|
| 685 |
+
"eval_runtime": 253.3553,
|
| 686 |
+
"eval_samples_per_second": 10.906,
|
| 687 |
+
"eval_steps_per_second": 1.366,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.3756298671552909,
|
| 692 |
+
"grad_norm": 1.2864240407943726,
|
| 693 |
+
"learning_rate": 7.8646021574225e-07,
|
| 694 |
+
"logits/chosen": -0.5721285343170166,
|
| 695 |
+
"logits/rejected": -0.5845418572425842,
|
| 696 |
+
"logps/chosen": -163.110595703125,
|
| 697 |
+
"logps/rejected": -217.90377807617188,
|
| 698 |
+
"loss": 0.3708,
|
| 699 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 700 |
+
"rewards/chosen": 2.0738914012908936,
|
| 701 |
+
"rewards/margins": 1.6767698526382446,
|
| 702 |
+
"rewards/rejected": 0.397121787071228,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.3847915712322492,
|
| 707 |
+
"grad_norm": 1.066874623298645,
|
| 708 |
+
"learning_rate": 7.731918090798113e-07,
|
| 709 |
+
"logits/chosen": -0.6550859212875366,
|
| 710 |
+
"logits/rejected": -0.6960932016372681,
|
| 711 |
+
"logps/chosen": -151.92108154296875,
|
| 712 |
+
"logps/rejected": -187.91384887695312,
|
| 713 |
+
"loss": 0.3284,
|
| 714 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 715 |
+
"rewards/chosen": 1.930013656616211,
|
| 716 |
+
"rewards/margins": 1.5344407558441162,
|
| 717 |
+
"rewards/rejected": 0.39557284116744995,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.39395327530920754,
|
| 722 |
+
"grad_norm": 1.4810634851455688,
|
| 723 |
+
"learning_rate": 7.596432513457482e-07,
|
| 724 |
+
"logits/chosen": -0.7274349927902222,
|
| 725 |
+
"logits/rejected": -0.7099400162696838,
|
| 726 |
+
"logps/chosen": -145.75698852539062,
|
| 727 |
+
"logps/rejected": -185.64678955078125,
|
| 728 |
+
"loss": 0.3359,
|
| 729 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 730 |
+
"rewards/chosen": 1.9275859594345093,
|
| 731 |
+
"rewards/margins": 1.4946165084838867,
|
| 732 |
+
"rewards/rejected": 0.43296942114830017,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.4031149793861658,
|
| 737 |
+
"grad_norm": 1.650376319885254,
|
| 738 |
+
"learning_rate": 7.458284362324842e-07,
|
| 739 |
+
"logits/chosen": -0.5438752174377441,
|
| 740 |
+
"logits/rejected": -0.6026032567024231,
|
| 741 |
+
"logps/chosen": -132.947021484375,
|
| 742 |
+
"logps/rejected": -219.47055053710938,
|
| 743 |
+
"loss": 0.3276,
|
| 744 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 745 |
+
"rewards/chosen": 1.8638055324554443,
|
| 746 |
+
"rewards/margins": 2.066131353378296,
|
| 747 |
+
"rewards/rejected": -0.20232591032981873,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.4122766834631241,
|
| 752 |
+
"grad_norm": 0.992289125919342,
|
| 753 |
+
"learning_rate": 7.317615304724387e-07,
|
| 754 |
+
"logits/chosen": -0.6501020193099976,
|
| 755 |
+
"logits/rejected": -0.680503249168396,
|
| 756 |
+
"logps/chosen": -157.24925231933594,
|
| 757 |
+
"logps/rejected": -180.8325958251953,
|
| 758 |
+
"loss": 0.3249,
|
| 759 |
+
"rewards/accuracies": 0.949999988079071,
|
| 760 |
+
"rewards/chosen": 1.7174543142318726,
|
| 761 |
+
"rewards/margins": 1.594322681427002,
|
| 762 |
+
"rewards/rejected": 0.12313172966241837,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.42143838754008245,
|
| 767 |
+
"grad_norm": 1.2098772525787354,
|
| 768 |
+
"learning_rate": 7.174569593104108e-07,
|
| 769 |
+
"logits/chosen": -0.7602720260620117,
|
| 770 |
+
"logits/rejected": -0.7695900201797485,
|
| 771 |
+
"logps/chosen": -174.65504455566406,
|
| 772 |
+
"logps/rejected": -225.5537567138672,
|
| 773 |
+
"loss": 0.3396,
|
| 774 |
+
"rewards/accuracies": 0.925000011920929,
|
| 775 |
+
"rewards/chosen": 1.8735862970352173,
|
| 776 |
+
"rewards/margins": 1.4029169082641602,
|
| 777 |
+
"rewards/rejected": 0.47066912055015564,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.43060009161704077,
|
| 782 |
+
"grad_norm": 1.896801471710205,
|
| 783 |
+
"learning_rate": 7.029293917108677e-07,
|
| 784 |
+
"logits/chosen": -0.6379343271255493,
|
| 785 |
+
"logits/rejected": -0.629081130027771,
|
| 786 |
+
"logps/chosen": -264.4695739746094,
|
| 787 |
+
"logps/rejected": -247.133056640625,
|
| 788 |
+
"loss": 0.3216,
|
| 789 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 790 |
+
"rewards/chosen": 1.6404482126235962,
|
| 791 |
+
"rewards/margins": 1.7144851684570312,
|
| 792 |
+
"rewards/rejected": -0.07403700053691864,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.4397617956939991,
|
| 797 |
+
"grad_norm": 2.521409273147583,
|
| 798 |
+
"learning_rate": 6.881937253153051e-07,
|
| 799 |
+
"logits/chosen": -0.7469202876091003,
|
| 800 |
+
"logits/rejected": -0.7621224522590637,
|
| 801 |
+
"logps/chosen": -165.92147827148438,
|
| 802 |
+
"logps/rejected": -223.8385772705078,
|
| 803 |
+
"loss": 0.3246,
|
| 804 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 805 |
+
"rewards/chosen": 2.0734469890594482,
|
| 806 |
+
"rewards/margins": 2.012188673019409,
|
| 807 |
+
"rewards/rejected": 0.061258465051651,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.4489234997709574,
|
| 812 |
+
"grad_norm": 1.2489752769470215,
|
| 813 |
+
"learning_rate": 6.732650711651031e-07,
|
| 814 |
+
"logits/chosen": -0.5696443319320679,
|
| 815 |
+
"logits/rejected": -0.618468165397644,
|
| 816 |
+
"logps/chosen": -189.3000030517578,
|
| 817 |
+
"logps/rejected": -246.3811798095703,
|
| 818 |
+
"loss": 0.2909,
|
| 819 |
+
"rewards/accuracies": 0.925000011920929,
|
| 820 |
+
"rewards/chosen": 1.7483714818954468,
|
| 821 |
+
"rewards/margins": 2.0019850730895996,
|
| 822 |
+
"rewards/rejected": -0.2536138892173767,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.45808520384791573,
|
| 827 |
+
"grad_norm": 1.3916680812835693,
|
| 828 |
+
"learning_rate": 6.581587382055491e-07,
|
| 829 |
+
"logits/chosen": -0.761835515499115,
|
| 830 |
+
"logits/rejected": -0.7702199220657349,
|
| 831 |
+
"logps/chosen": -151.83370971679688,
|
| 832 |
+
"logps/rejected": -229.975341796875,
|
| 833 |
+
"loss": 0.3064,
|
| 834 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 835 |
+
"rewards/chosen": 1.8061788082122803,
|
| 836 |
+
"rewards/margins": 1.7005802392959595,
|
| 837 |
+
"rewards/rejected": 0.10559873282909393,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.45808520384791573,
|
| 842 |
+
"eval_logits/chosen": -0.6487900018692017,
|
| 843 |
+
"eval_logits/rejected": -0.6642398834228516,
|
| 844 |
+
"eval_logps/chosen": -199.9703826904297,
|
| 845 |
+
"eval_logps/rejected": -267.3388671875,
|
| 846 |
+
"eval_loss": 0.347669780254364,
|
| 847 |
+
"eval_rewards/accuracies": 0.8612716794013977,
|
| 848 |
+
"eval_rewards/chosen": 1.7220263481140137,
|
| 849 |
+
"eval_rewards/margins": 1.7402905225753784,
|
| 850 |
+
"eval_rewards/rejected": -0.01826408877968788,
|
| 851 |
+
"eval_runtime": 253.4005,
|
| 852 |
+
"eval_samples_per_second": 10.904,
|
| 853 |
+
"eval_steps_per_second": 1.365,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.467246907924874,
|
| 858 |
+
"grad_norm": 1.3511228561401367,
|
| 859 |
+
"learning_rate": 6.428902175869126e-07,
|
| 860 |
+
"logits/chosen": -0.680508553981781,
|
| 861 |
+
"logits/rejected": -0.6774734258651733,
|
| 862 |
+
"logps/chosen": -177.41610717773438,
|
| 863 |
+
"logps/rejected": -221.32138061523438,
|
| 864 |
+
"loss": 0.3258,
|
| 865 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 866 |
+
"rewards/chosen": 2.022366762161255,
|
| 867 |
+
"rewards/margins": 2.05668044090271,
|
| 868 |
+
"rewards/rejected": -0.03431398794054985,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.4764086120018323,
|
| 873 |
+
"grad_norm": 0.9841328263282776,
|
| 874 |
+
"learning_rate": 6.274751667786761e-07,
|
| 875 |
+
"logits/chosen": -0.6339150667190552,
|
| 876 |
+
"logits/rejected": -0.5782631635665894,
|
| 877 |
+
"logps/chosen": -230.2734832763672,
|
| 878 |
+
"logps/rejected": -311.7682800292969,
|
| 879 |
+
"loss": 0.3229,
|
| 880 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 881 |
+
"rewards/chosen": 1.7927030324935913,
|
| 882 |
+
"rewards/margins": 1.7812907695770264,
|
| 883 |
+
"rewards/rejected": 0.011412340216338634,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.48557031607879064,
|
| 888 |
+
"grad_norm": 2.7643489837646484,
|
| 889 |
+
"learning_rate": 6.119293935132075e-07,
|
| 890 |
+
"logits/chosen": -0.6203776597976685,
|
| 891 |
+
"logits/rejected": -0.6682835817337036,
|
| 892 |
+
"logps/chosen": -151.23324584960938,
|
| 893 |
+
"logps/rejected": -187.77114868164062,
|
| 894 |
+
"loss": 0.3034,
|
| 895 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 896 |
+
"rewards/chosen": 1.8519824743270874,
|
| 897 |
+
"rewards/margins": 2.0718624591827393,
|
| 898 |
+
"rewards/rejected": -0.21987971663475037,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.49473202015574896,
|
| 903 |
+
"grad_norm": 1.3088020086288452,
|
| 904 |
+
"learning_rate": 5.962688395753437e-07,
|
| 905 |
+
"logits/chosen": -0.8648042678833008,
|
| 906 |
+
"logits/rejected": -0.9114105105400085,
|
| 907 |
+
"logps/chosen": -137.6197052001953,
|
| 908 |
+
"logps/rejected": -205.31900024414062,
|
| 909 |
+
"loss": 0.2938,
|
| 910 |
+
"rewards/accuracies": 0.949999988079071,
|
| 911 |
+
"rewards/chosen": 1.5214287042617798,
|
| 912 |
+
"rewards/margins": 1.6869144439697266,
|
| 913 |
+
"rewards/rejected": -0.16548602283000946,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.5038937242327073,
|
| 918 |
+
"grad_norm": 1.9471220970153809,
|
| 919 |
+
"learning_rate": 5.80509564454506e-07,
|
| 920 |
+
"logits/chosen": -0.7066371440887451,
|
| 921 |
+
"logits/rejected": -0.6964636445045471,
|
| 922 |
+
"logps/chosen": -106.696533203125,
|
| 923 |
+
"logps/rejected": -200.95729064941406,
|
| 924 |
+
"loss": 0.311,
|
| 925 |
+
"rewards/accuracies": 0.824999988079071,
|
| 926 |
+
"rewards/chosen": 1.6337581872940063,
|
| 927 |
+
"rewards/margins": 1.6309497356414795,
|
| 928 |
+
"rewards/rejected": 0.0028083801735192537,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.5130554283096656,
|
| 933 |
+
"grad_norm": 1.3545409440994263,
|
| 934 |
+
"learning_rate": 5.646677288761132e-07,
|
| 935 |
+
"logits/chosen": -0.653856098651886,
|
| 936 |
+
"logits/rejected": -0.7129830121994019,
|
| 937 |
+
"logps/chosen": -153.89035034179688,
|
| 938 |
+
"logps/rejected": -214.0993194580078,
|
| 939 |
+
"loss": 0.3112,
|
| 940 |
+
"rewards/accuracies": 0.949999988079071,
|
| 941 |
+
"rewards/chosen": 1.5801527500152588,
|
| 942 |
+
"rewards/margins": 1.731884241104126,
|
| 943 |
+
"rewards/rejected": -0.1517314463853836,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.5222171323866239,
|
| 948 |
+
"grad_norm": 2.128307819366455,
|
| 949 |
+
"learning_rate": 5.487595782291784e-07,
|
| 950 |
+
"logits/chosen": -0.6990654468536377,
|
| 951 |
+
"logits/rejected": -0.7342058420181274,
|
| 952 |
+
"logps/chosen": -178.05380249023438,
|
| 953 |
+
"logps/rejected": -225.2575225830078,
|
| 954 |
+
"loss": 0.289,
|
| 955 |
+
"rewards/accuracies": 0.824999988079071,
|
| 956 |
+
"rewards/chosen": 1.7597728967666626,
|
| 957 |
+
"rewards/margins": 1.7312393188476562,
|
| 958 |
+
"rewards/rejected": 0.02853367291390896,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.5313788364635822,
|
| 963 |
+
"grad_norm": 1.2450644969940186,
|
| 964 |
+
"learning_rate": 5.328014259070878e-07,
|
| 965 |
+
"logits/chosen": -0.619064211845398,
|
| 966 |
+
"logits/rejected": -0.6336346864700317,
|
| 967 |
+
"logps/chosen": -188.67752075195312,
|
| 968 |
+
"logps/rejected": -231.69503784179688,
|
| 969 |
+
"loss": 0.3388,
|
| 970 |
+
"rewards/accuracies": 0.925000011920929,
|
| 971 |
+
"rewards/chosen": 1.4915143251419067,
|
| 972 |
+
"rewards/margins": 1.5344158411026,
|
| 973 |
+
"rewards/rejected": -0.042901456356048584,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.5405405405405406,
|
| 978 |
+
"grad_norm": 1.2011312246322632,
|
| 979 |
+
"learning_rate": 5.168096365786402e-07,
|
| 980 |
+
"logits/chosen": -0.7081841230392456,
|
| 981 |
+
"logits/rejected": -0.7206417918205261,
|
| 982 |
+
"logps/chosen": -163.11752319335938,
|
| 983 |
+
"logps/rejected": -235.26065063476562,
|
| 984 |
+
"loss": 0.311,
|
| 985 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 986 |
+
"rewards/chosen": 1.9438568353652954,
|
| 987 |
+
"rewards/margins": 2.347649574279785,
|
| 988 |
+
"rewards/rejected": -0.4037927985191345,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.5497022446174988,
|
| 993 |
+
"grad_norm": 1.7624154090881348,
|
| 994 |
+
"learning_rate": 5.008006094065069e-07,
|
| 995 |
+
"logits/chosen": -0.721166729927063,
|
| 996 |
+
"logits/rejected": -0.7850608229637146,
|
| 997 |
+
"logps/chosen": -165.5583953857422,
|
| 998 |
+
"logps/rejected": -213.85855102539062,
|
| 999 |
+
"loss": 0.3054,
|
| 1000 |
+
"rewards/accuracies": 0.875,
|
| 1001 |
+
"rewards/chosen": 1.5962111949920654,
|
| 1002 |
+
"rewards/margins": 1.4221036434173584,
|
| 1003 |
+
"rewards/rejected": 0.174107626080513,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.5497022446174988,
|
| 1008 |
+
"eval_logits/chosen": -0.6406525373458862,
|
| 1009 |
+
"eval_logits/rejected": -0.6576036214828491,
|
| 1010 |
+
"eval_logps/chosen": -201.4723358154297,
|
| 1011 |
+
"eval_logps/rejected": -270.9281311035156,
|
| 1012 |
+
"eval_loss": 0.3270590603351593,
|
| 1013 |
+
"eval_rewards/accuracies": 0.8670520186424255,
|
| 1014 |
+
"eval_rewards/chosen": 1.6469277143478394,
|
| 1015 |
+
"eval_rewards/margins": 1.8446547985076904,
|
| 1016 |
+
"eval_rewards/rejected": -0.1977270245552063,
|
| 1017 |
+
"eval_runtime": 253.7625,
|
| 1018 |
+
"eval_samples_per_second": 10.888,
|
| 1019 |
+
"eval_steps_per_second": 1.363,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.5588639486944572,
|
| 1024 |
+
"grad_norm": 1.902051329612732,
|
| 1025 |
+
"learning_rate": 4.847907612303182e-07,
|
| 1026 |
+
"logits/chosen": -0.7130570411682129,
|
| 1027 |
+
"logits/rejected": -0.7378814816474915,
|
| 1028 |
+
"logps/chosen": -188.9144287109375,
|
| 1029 |
+
"logps/rejected": -259.7423400878906,
|
| 1030 |
+
"loss": 0.3009,
|
| 1031 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1032 |
+
"rewards/chosen": 1.5585817098617554,
|
| 1033 |
+
"rewards/margins": 1.3624755144119263,
|
| 1034 |
+
"rewards/rejected": 0.19610631465911865,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.5680256527714155,
|
| 1039 |
+
"grad_norm": 1.07755446434021,
|
| 1040 |
+
"learning_rate": 4.687965097316223e-07,
|
| 1041 |
+
"logits/chosen": -0.5912365317344666,
|
| 1042 |
+
"logits/rejected": -0.7409440875053406,
|
| 1043 |
+
"logps/chosen": -126.41873931884766,
|
| 1044 |
+
"logps/rejected": -238.76437377929688,
|
| 1045 |
+
"loss": 0.275,
|
| 1046 |
+
"rewards/accuracies": 0.875,
|
| 1047 |
+
"rewards/chosen": 1.855369210243225,
|
| 1048 |
+
"rewards/margins": 2.4938321113586426,
|
| 1049 |
+
"rewards/rejected": -0.638463020324707,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.5771873568483737,
|
| 1054 |
+
"grad_norm": 1.3470137119293213,
|
| 1055 |
+
"learning_rate": 4.5283425659798175e-07,
|
| 1056 |
+
"logits/chosen": -0.8164669275283813,
|
| 1057 |
+
"logits/rejected": -0.8073943853378296,
|
| 1058 |
+
"logps/chosen": -201.78501892089844,
|
| 1059 |
+
"logps/rejected": -291.32183837890625,
|
| 1060 |
+
"loss": 0.3179,
|
| 1061 |
+
"rewards/accuracies": 0.875,
|
| 1062 |
+
"rewards/chosen": 1.852560043334961,
|
| 1063 |
+
"rewards/margins": 1.8420225381851196,
|
| 1064 |
+
"rewards/rejected": 0.010537643916904926,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.5863490609253321,
|
| 1069 |
+
"grad_norm": 1.2435747385025024,
|
| 1070 |
+
"learning_rate": 4.3692037070347123e-07,
|
| 1071 |
+
"logits/chosen": -0.6459102630615234,
|
| 1072 |
+
"logits/rejected": -0.6654535531997681,
|
| 1073 |
+
"logps/chosen": -139.29067993164062,
|
| 1074 |
+
"logps/rejected": -212.83740234375,
|
| 1075 |
+
"loss": 0.3016,
|
| 1076 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1077 |
+
"rewards/chosen": 1.6537023782730103,
|
| 1078 |
+
"rewards/margins": 2.188789129257202,
|
| 1079 |
+
"rewards/rejected": -0.5350866913795471,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.5955107650022904,
|
| 1084 |
+
"grad_norm": 1.0110821723937988,
|
| 1085 |
+
"learning_rate": 4.21071171322823e-07,
|
| 1086 |
+
"logits/chosen": -0.6093601584434509,
|
| 1087 |
+
"logits/rejected": -0.5892384648323059,
|
| 1088 |
+
"logps/chosen": -275.14105224609375,
|
| 1089 |
+
"logps/rejected": -322.5516662597656,
|
| 1090 |
+
"loss": 0.3134,
|
| 1091 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1092 |
+
"rewards/chosen": 1.5108485221862793,
|
| 1093 |
+
"rewards/margins": 1.933189034461975,
|
| 1094 |
+
"rewards/rejected": -0.4223404824733734,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6046724690792488,
|
| 1099 |
+
"grad_norm": 0.9678570628166199,
|
| 1100 |
+
"learning_rate": 4.0530291139643755e-07,
|
| 1101 |
+
"logits/chosen": -0.8226197957992554,
|
| 1102 |
+
"logits/rejected": -0.8122960925102234,
|
| 1103 |
+
"logps/chosen": -143.47821044921875,
|
| 1104 |
+
"logps/rejected": -207.33187866210938,
|
| 1105 |
+
"loss": 0.2829,
|
| 1106 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1107 |
+
"rewards/chosen": 2.0050506591796875,
|
| 1108 |
+
"rewards/margins": 2.1615443229675293,
|
| 1109 |
+
"rewards/rejected": -0.1564939320087433,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.613834173156207,
|
| 1114 |
+
"grad_norm": 1.5877900123596191,
|
| 1115 |
+
"learning_rate": 3.8963176086341727e-07,
|
| 1116 |
+
"logits/chosen": -0.6404609084129333,
|
| 1117 |
+
"logits/rejected": -0.7193113565444946,
|
| 1118 |
+
"logps/chosen": -163.66485595703125,
|
| 1119 |
+
"logps/rejected": -219.49063110351562,
|
| 1120 |
+
"loss": 0.2704,
|
| 1121 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1122 |
+
"rewards/chosen": 1.613854169845581,
|
| 1123 |
+
"rewards/margins": 2.1832098960876465,
|
| 1124 |
+
"rewards/rejected": -0.5693557858467102,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.6229958772331654,
|
| 1129 |
+
"grad_norm": 1.301003098487854,
|
| 1130 |
+
"learning_rate": 3.7407379007971506e-07,
|
| 1131 |
+
"logits/chosen": -0.6765211820602417,
|
| 1132 |
+
"logits/rejected": -0.6541970372200012,
|
| 1133 |
+
"logps/chosen": -199.01051330566406,
|
| 1134 |
+
"logps/rejected": -279.47454833984375,
|
| 1135 |
+
"loss": 0.2896,
|
| 1136 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1137 |
+
"rewards/chosen": 1.5939905643463135,
|
| 1138 |
+
"rewards/margins": 2.2719438076019287,
|
| 1139 |
+
"rewards/rejected": -0.6779531240463257,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.6321575813101237,
|
| 1144 |
+
"grad_norm": 2.2288615703582764,
|
| 1145 |
+
"learning_rate": 3.586449533384048e-07,
|
| 1146 |
+
"logits/chosen": -0.6064814925193787,
|
| 1147 |
+
"logits/rejected": -0.6105560660362244,
|
| 1148 |
+
"logps/chosen": -137.80624389648438,
|
| 1149 |
+
"logps/rejected": -186.99191284179688,
|
| 1150 |
+
"loss": 0.3041,
|
| 1151 |
+
"rewards/accuracies": 0.875,
|
| 1152 |
+
"rewards/chosen": 1.5289627313613892,
|
| 1153 |
+
"rewards/margins": 1.8546040058135986,
|
| 1154 |
+
"rewards/rejected": -0.32564133405685425,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.641319285387082,
|
| 1159 |
+
"grad_norm": 5.320913314819336,
|
| 1160 |
+
"learning_rate": 3.433610725089692e-07,
|
| 1161 |
+
"logits/chosen": -0.7031580209732056,
|
| 1162 |
+
"logits/rejected": -0.6630051136016846,
|
| 1163 |
+
"logps/chosen": -185.18197631835938,
|
| 1164 |
+
"logps/rejected": -280.9058532714844,
|
| 1165 |
+
"loss": 0.2919,
|
| 1166 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1167 |
+
"rewards/chosen": 1.8430522680282593,
|
| 1168 |
+
"rewards/margins": 1.9076077938079834,
|
| 1169 |
+
"rewards/rejected": -0.06455531716346741,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.641319285387082,
|
| 1174 |
+
"eval_logits/chosen": -0.6672143936157227,
|
| 1175 |
+
"eval_logits/rejected": -0.6753049492835999,
|
| 1176 |
+
"eval_logps/chosen": -199.65896606445312,
|
| 1177 |
+
"eval_logps/rejected": -273.0414123535156,
|
| 1178 |
+
"eval_loss": 0.3144252896308899,
|
| 1179 |
+
"eval_rewards/accuracies": 0.8641618490219116,
|
| 1180 |
+
"eval_rewards/chosen": 1.7375967502593994,
|
| 1181 |
+
"eval_rewards/margins": 2.0409882068634033,
|
| 1182 |
+
"eval_rewards/rejected": -0.30339136719703674,
|
| 1183 |
+
"eval_runtime": 253.1657,
|
| 1184 |
+
"eval_samples_per_second": 10.914,
|
| 1185 |
+
"eval_steps_per_second": 1.367,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.6504809894640403,
|
| 1190 |
+
"grad_norm": 1.6359617710113525,
|
| 1191 |
+
"learning_rate": 3.2823782081238555e-07,
|
| 1192 |
+
"logits/chosen": -0.7354801893234253,
|
| 1193 |
+
"logits/rejected": -0.7596527338027954,
|
| 1194 |
+
"logps/chosen": -145.60165405273438,
|
| 1195 |
+
"logps/rejected": -207.4108123779297,
|
| 1196 |
+
"loss": 0.3059,
|
| 1197 |
+
"rewards/accuracies": 0.875,
|
| 1198 |
+
"rewards/chosen": 1.8920338153839111,
|
| 1199 |
+
"rewards/margins": 1.903387427330017,
|
| 1200 |
+
"rewards/rejected": -0.011353528127074242,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.6596426935409986,
|
| 1205 |
+
"grad_norm": 1.6316208839416504,
|
| 1206 |
+
"learning_rate": 3.132907067486471e-07,
|
| 1207 |
+
"logits/chosen": -0.6749696135520935,
|
| 1208 |
+
"logits/rejected": -0.7063171863555908,
|
| 1209 |
+
"logps/chosen": -158.86273193359375,
|
| 1210 |
+
"logps/rejected": -211.850341796875,
|
| 1211 |
+
"loss": 0.2837,
|
| 1212 |
+
"rewards/accuracies": 0.875,
|
| 1213 |
+
"rewards/chosen": 1.6359964609146118,
|
| 1214 |
+
"rewards/margins": 1.919615387916565,
|
| 1215 |
+
"rewards/rejected": -0.2836189270019531,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.668804397617957,
|
| 1220 |
+
"grad_norm": 2.493192672729492,
|
| 1221 |
+
"learning_rate": 2.985350581932005e-07,
|
| 1222 |
+
"logits/chosen": -0.7640320062637329,
|
| 1223 |
+
"logits/rejected": -0.7546281814575195,
|
| 1224 |
+
"logps/chosen": -185.2474822998047,
|
| 1225 |
+
"logps/rejected": -231.47119140625,
|
| 1226 |
+
"loss": 0.3068,
|
| 1227 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1228 |
+
"rewards/chosen": 2.423882007598877,
|
| 1229 |
+
"rewards/margins": 2.5569610595703125,
|
| 1230 |
+
"rewards/rejected": -0.13307929039001465,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 0.6779661016949152,
|
| 1235 |
+
"grad_norm": 1.0160990953445435,
|
| 1236 |
+
"learning_rate": 2.839860066786103e-07,
|
| 1237 |
+
"logits/chosen": -0.7123746871948242,
|
| 1238 |
+
"logits/rejected": -0.7270351648330688,
|
| 1239 |
+
"logps/chosen": -131.58505249023438,
|
| 1240 |
+
"logps/rejected": -189.05264282226562,
|
| 1241 |
+
"loss": 0.2778,
|
| 1242 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1243 |
+
"rewards/chosen": 1.8099651336669922,
|
| 1244 |
+
"rewards/margins": 2.3398964405059814,
|
| 1245 |
+
"rewards/rejected": -0.5299314260482788,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 0.6871278057718736,
|
| 1250 |
+
"grad_norm": 1.4675796031951904,
|
| 1251 |
+
"learning_rate": 2.6965847187756553e-07,
|
| 1252 |
+
"logits/chosen": -0.7339153289794922,
|
| 1253 |
+
"logits/rejected": -0.6953274011611938,
|
| 1254 |
+
"logps/chosen": -144.21389770507812,
|
| 1255 |
+
"logps/rejected": -171.4956512451172,
|
| 1256 |
+
"loss": 0.2746,
|
| 1257 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1258 |
+
"rewards/chosen": 2.1637752056121826,
|
| 1259 |
+
"rewards/margins": 2.3990566730499268,
|
| 1260 |
+
"rewards/rejected": -0.2352815419435501,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.6962895098488319,
|
| 1265 |
+
"grad_norm": 2.0078845024108887,
|
| 1266 |
+
"learning_rate": 2.5556714630314613e-07,
|
| 1267 |
+
"logits/chosen": -0.783401370048523,
|
| 1268 |
+
"logits/rejected": -0.854455828666687,
|
| 1269 |
+
"logps/chosen": -105.17523193359375,
|
| 1270 |
+
"logps/rejected": -193.60110473632812,
|
| 1271 |
+
"loss": 0.2727,
|
| 1272 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1273 |
+
"rewards/chosen": 2.012260913848877,
|
| 1274 |
+
"rewards/margins": 2.8885557651519775,
|
| 1275 |
+
"rewards/rejected": -0.8762944340705872,
|
| 1276 |
+
"step": 760
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.7054512139257902,
|
| 1280 |
+
"grad_norm": 1.2648169994354248,
|
| 1281 |
+
"learning_rate": 2.417264802420343e-07,
|
| 1282 |
+
"logits/chosen": -0.6513649225234985,
|
| 1283 |
+
"logits/rejected": -0.6898313760757446,
|
| 1284 |
+
"logps/chosen": -141.83282470703125,
|
| 1285 |
+
"logps/rejected": -238.6311492919922,
|
| 1286 |
+
"loss": 0.3154,
|
| 1287 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1288 |
+
"rewards/chosen": 1.7510058879852295,
|
| 1289 |
+
"rewards/margins": 2.0821216106414795,
|
| 1290 |
+
"rewards/rejected": -0.33111587166786194,
|
| 1291 |
+
"step": 770
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.7146129180027485,
|
| 1295 |
+
"grad_norm": 2.6904611587524414,
|
| 1296 |
+
"learning_rate": 2.2815066693612117e-07,
|
| 1297 |
+
"logits/chosen": -0.7128955721855164,
|
| 1298 |
+
"logits/rejected": -0.7633499503135681,
|
| 1299 |
+
"logps/chosen": -155.6310577392578,
|
| 1300 |
+
"logps/rejected": -210.39285278320312,
|
| 1301 |
+
"loss": 0.2793,
|
| 1302 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1303 |
+
"rewards/chosen": 1.7025461196899414,
|
| 1304 |
+
"rewards/margins": 2.1338300704956055,
|
| 1305 |
+
"rewards/rejected": -0.4312838613986969,
|
| 1306 |
+
"step": 780
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.7237746220797068,
|
| 1310 |
+
"grad_norm": 2.1170003414154053,
|
| 1311 |
+
"learning_rate": 2.1485362802770862e-07,
|
| 1312 |
+
"logits/chosen": -0.6805752515792847,
|
| 1313 |
+
"logits/rejected": -0.7278204560279846,
|
| 1314 |
+
"logps/chosen": -207.82858276367188,
|
| 1315 |
+
"logps/rejected": -320.7551574707031,
|
| 1316 |
+
"loss": 0.2457,
|
| 1317 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1318 |
+
"rewards/chosen": 1.8095613718032837,
|
| 1319 |
+
"rewards/margins": 2.516162157058716,
|
| 1320 |
+
"rewards/rejected": -0.7066007852554321,
|
| 1321 |
+
"step": 790
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.7329363261566652,
|
| 1325 |
+
"grad_norm": 1.6859688758850098,
|
| 1326 |
+
"learning_rate": 2.018489992832283e-07,
|
| 1327 |
+
"logits/chosen": -0.7043607831001282,
|
| 1328 |
+
"logits/rejected": -0.6326649785041809,
|
| 1329 |
+
"logps/chosen": -189.71005249023438,
|
| 1330 |
+
"logps/rejected": -247.50723266601562,
|
| 1331 |
+
"loss": 0.314,
|
| 1332 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1333 |
+
"rewards/chosen": 1.972821593284607,
|
| 1334 |
+
"rewards/margins": 2.5004470348358154,
|
| 1335 |
+
"rewards/rejected": -0.5276254415512085,
|
| 1336 |
+
"step": 800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 0.7329363261566652,
|
| 1340 |
+
"eval_logits/chosen": -0.6573572754859924,
|
| 1341 |
+
"eval_logits/rejected": -0.6684801578521729,
|
| 1342 |
+
"eval_logps/chosen": -200.33786010742188,
|
| 1343 |
+
"eval_logps/rejected": -275.4322814941406,
|
| 1344 |
+
"eval_loss": 0.3056153357028961,
|
| 1345 |
+
"eval_rewards/accuracies": 0.8670520186424255,
|
| 1346 |
+
"eval_rewards/chosen": 1.703651785850525,
|
| 1347 |
+
"eval_rewards/margins": 2.126586437225342,
|
| 1348 |
+
"eval_rewards/rejected": -0.4229348599910736,
|
| 1349 |
+
"eval_runtime": 253.4624,
|
| 1350 |
+
"eval_samples_per_second": 10.901,
|
| 1351 |
+
"eval_steps_per_second": 1.365,
|
| 1352 |
+
"step": 800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.7420980302336234,
|
| 1356 |
+
"grad_norm": 1.46702241897583,
|
| 1357 |
+
"learning_rate": 1.891501166101187e-07,
|
| 1358 |
+
"logits/chosen": -0.8299296498298645,
|
| 1359 |
+
"logits/rejected": -0.8194143176078796,
|
| 1360 |
+
"logps/chosen": -139.70791625976562,
|
| 1361 |
+
"logps/rejected": -179.2274932861328,
|
| 1362 |
+
"loss": 0.2988,
|
| 1363 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1364 |
+
"rewards/chosen": 1.740025281906128,
|
| 1365 |
+
"rewards/margins": 2.0235095024108887,
|
| 1366 |
+
"rewards/rejected": -0.28348422050476074,
|
| 1367 |
+
"step": 810
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.7512597343105818,
|
| 1371 |
+
"grad_norm": 2.791546583175659,
|
| 1372 |
+
"learning_rate": 1.767700023812e-07,
|
| 1373 |
+
"logits/chosen": -0.7000871896743774,
|
| 1374 |
+
"logits/rejected": -0.7028741240501404,
|
| 1375 |
+
"logps/chosen": -174.82821655273438,
|
| 1376 |
+
"logps/rejected": -264.07470703125,
|
| 1377 |
+
"loss": 0.272,
|
| 1378 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1379 |
+
"rewards/chosen": 2.0946412086486816,
|
| 1380 |
+
"rewards/margins": 2.783123016357422,
|
| 1381 |
+
"rewards/rejected": -0.6884818077087402,
|
| 1382 |
+
"step": 820
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.7604214383875401,
|
| 1386 |
+
"grad_norm": 1.3349977731704712,
|
| 1387 |
+
"learning_rate": 1.6472135208057125e-07,
|
| 1388 |
+
"logits/chosen": -0.6306108832359314,
|
| 1389 |
+
"logits/rejected": -0.698475182056427,
|
| 1390 |
+
"logps/chosen": -155.9135284423828,
|
| 1391 |
+
"logps/rejected": -198.2060546875,
|
| 1392 |
+
"loss": 0.3032,
|
| 1393 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1394 |
+
"rewards/chosen": 1.55352783203125,
|
| 1395 |
+
"rewards/margins": 2.126960515975952,
|
| 1396 |
+
"rewards/rejected": -0.5734325647354126,
|
| 1397 |
+
"step": 830
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.7695831424644984,
|
| 1401 |
+
"grad_norm": 2.1777162551879883,
|
| 1402 |
+
"learning_rate": 1.530165212847217e-07,
|
| 1403 |
+
"logits/chosen": -0.7774958610534668,
|
| 1404 |
+
"logits/rejected": -0.7312562465667725,
|
| 1405 |
+
"logps/chosen": -141.79075622558594,
|
| 1406 |
+
"logps/rejected": -190.49310302734375,
|
| 1407 |
+
"loss": 0.2807,
|
| 1408 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1409 |
+
"rewards/chosen": 1.8885326385498047,
|
| 1410 |
+
"rewards/margins": 2.229785442352295,
|
| 1411 |
+
"rewards/rejected": -0.3412528932094574,
|
| 1412 |
+
"step": 840
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.7787448465414567,
|
| 1416 |
+
"grad_norm": 1.2950890064239502,
|
| 1417 |
+
"learning_rate": 1.4166751299221003e-07,
|
| 1418 |
+
"logits/chosen": -0.6935632228851318,
|
| 1419 |
+
"logits/rejected": -0.681289553642273,
|
| 1420 |
+
"logps/chosen": -166.78561401367188,
|
| 1421 |
+
"logps/rejected": -229.1016387939453,
|
| 1422 |
+
"loss": 0.2802,
|
| 1423 |
+
"rewards/accuracies": 0.875,
|
| 1424 |
+
"rewards/chosen": 1.8389393091201782,
|
| 1425 |
+
"rewards/margins": 3.066588878631592,
|
| 1426 |
+
"rewards/rejected": -1.2276496887207031,
|
| 1427 |
+
"step": 850
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.7879065506184151,
|
| 1431 |
+
"grad_norm": 1.1921981573104858,
|
| 1432 |
+
"learning_rate": 1.306859653149025e-07,
|
| 1433 |
+
"logits/chosen": -0.7318634390830994,
|
| 1434 |
+
"logits/rejected": -0.7256627082824707,
|
| 1435 |
+
"logps/chosen": -171.42092895507812,
|
| 1436 |
+
"logps/rejected": -247.11831665039062,
|
| 1437 |
+
"loss": 0.2934,
|
| 1438 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1439 |
+
"rewards/chosen": 1.7239230871200562,
|
| 1440 |
+
"rewards/margins": 2.5621562004089355,
|
| 1441 |
+
"rewards/rejected": -0.8382335901260376,
|
| 1442 |
+
"step": 860
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.7970682546953733,
|
| 1446 |
+
"grad_norm": 1.4422613382339478,
|
| 1447 |
+
"learning_rate": 1.2008313954339305e-07,
|
| 1448 |
+
"logits/chosen": -0.586450457572937,
|
| 1449 |
+
"logits/rejected": -0.5815967321395874,
|
| 1450 |
+
"logps/chosen": -215.48007202148438,
|
| 1451 |
+
"logps/rejected": -242.58511352539062,
|
| 1452 |
+
"loss": 0.2549,
|
| 1453 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1454 |
+
"rewards/chosen": 1.6724342107772827,
|
| 1455 |
+
"rewards/margins": 1.982318639755249,
|
| 1456 |
+
"rewards/rejected": -0.3098844885826111,
|
| 1457 |
+
"step": 870
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.8062299587723316,
|
| 1461 |
+
"grad_norm": 1.141564130783081,
|
| 1462 |
+
"learning_rate": 1.098699085988432e-07,
|
| 1463 |
+
"logits/chosen": -0.7615233659744263,
|
| 1464 |
+
"logits/rejected": -0.8405082821846008,
|
| 1465 |
+
"logps/chosen": -163.58689880371094,
|
| 1466 |
+
"logps/rejected": -256.6318054199219,
|
| 1467 |
+
"loss": 0.2998,
|
| 1468 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1469 |
+
"rewards/chosen": 1.7078421115875244,
|
| 1470 |
+
"rewards/margins": 1.9117094278335571,
|
| 1471 |
+
"rewards/rejected": -0.20386750996112823,
|
| 1472 |
+
"step": 880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.81539166284929,
|
| 1476 |
+
"grad_norm": 1.329354166984558,
|
| 1477 |
+
"learning_rate": 1.0005674588308566e-07,
|
| 1478 |
+
"logits/chosen": -0.7380274534225464,
|
| 1479 |
+
"logits/rejected": -0.75429767370224,
|
| 1480 |
+
"logps/chosen": -136.90902709960938,
|
| 1481 |
+
"logps/rejected": -224.83151245117188,
|
| 1482 |
+
"loss": 0.2231,
|
| 1483 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1484 |
+
"rewards/chosen": 1.7314188480377197,
|
| 1485 |
+
"rewards/margins": 2.723328113555908,
|
| 1486 |
+
"rewards/rejected": -0.9919096231460571,
|
| 1487 |
+
"step": 890
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.8245533669262483,
|
| 1491 |
+
"grad_norm": 1.717936396598816,
|
| 1492 |
+
"learning_rate": 9.065371453842358e-08,
|
| 1493 |
+
"logits/chosen": -0.670835018157959,
|
| 1494 |
+
"logits/rejected": -0.6879727244377136,
|
| 1495 |
+
"logps/chosen": -138.87889099121094,
|
| 1496 |
+
"logps/rejected": -184.62257385253906,
|
| 1497 |
+
"loss": 0.3014,
|
| 1498 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1499 |
+
"rewards/chosen": 2.0553038120269775,
|
| 1500 |
+
"rewards/margins": 2.568469524383545,
|
| 1501 |
+
"rewards/rejected": -0.5131659507751465,
|
| 1502 |
+
"step": 900
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.8245533669262483,
|
| 1506 |
+
"eval_logits/chosen": -0.6641213893890381,
|
| 1507 |
+
"eval_logits/rejected": -0.670166015625,
|
| 1508 |
+
"eval_logps/chosen": -200.79710388183594,
|
| 1509 |
+
"eval_logps/rejected": -276.2374267578125,
|
| 1510 |
+
"eval_loss": 0.3019951581954956,
|
| 1511 |
+
"eval_rewards/accuracies": 0.8699421882629395,
|
| 1512 |
+
"eval_rewards/chosen": 1.6806890964508057,
|
| 1513 |
+
"eval_rewards/margins": 2.1438791751861572,
|
| 1514 |
+
"eval_rewards/rejected": -0.4631901979446411,
|
| 1515 |
+
"eval_runtime": 253.6623,
|
| 1516 |
+
"eval_samples_per_second": 10.892,
|
| 1517 |
+
"eval_steps_per_second": 1.364,
|
| 1518 |
+
"step": 900
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 0.8337150710032066,
|
| 1522 |
+
"grad_norm": 1.479974627494812,
|
| 1523 |
+
"learning_rate": 8.167045712814108e-08,
|
| 1524 |
+
"logits/chosen": -0.6214216351509094,
|
| 1525 |
+
"logits/rejected": -0.641666054725647,
|
| 1526 |
+
"logps/chosen": -177.58828735351562,
|
| 1527 |
+
"logps/rejected": -273.30755615234375,
|
| 1528 |
+
"loss": 0.2533,
|
| 1529 |
+
"rewards/accuracies": 0.875,
|
| 1530 |
+
"rewards/chosen": 1.7214215993881226,
|
| 1531 |
+
"rewards/margins": 1.9738502502441406,
|
| 1532 |
+
"rewards/rejected": -0.2524286210536957,
|
| 1533 |
+
"step": 910
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.8428767750801649,
|
| 1537 |
+
"grad_norm": 2.062375783920288,
|
| 1538 |
+
"learning_rate": 7.311618574830569e-08,
|
| 1539 |
+
"logits/chosen": -0.5943703651428223,
|
| 1540 |
+
"logits/rejected": -0.6128894686698914,
|
| 1541 |
+
"logps/chosen": -163.90029907226562,
|
| 1542 |
+
"logps/rejected": -248.54415893554688,
|
| 1543 |
+
"loss": 0.2868,
|
| 1544 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1545 |
+
"rewards/chosen": 2.0122509002685547,
|
| 1546 |
+
"rewards/margins": 2.265479803085327,
|
| 1547 |
+
"rewards/rejected": -0.25322893261909485,
|
| 1548 |
+
"step": 920
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 0.8520384791571233,
|
| 1552 |
+
"grad_norm": 1.7314789295196533,
|
| 1553 |
+
"learning_rate": 6.499967258100514e-08,
|
| 1554 |
+
"logits/chosen": -0.6944621801376343,
|
| 1555 |
+
"logits/rejected": -0.7938731908798218,
|
| 1556 |
+
"logps/chosen": -163.20828247070312,
|
| 1557 |
+
"logps/rejected": -237.55078125,
|
| 1558 |
+
"loss": 0.2645,
|
| 1559 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1560 |
+
"rewards/chosen": 1.7433500289916992,
|
| 1561 |
+
"rewards/margins": 2.302280902862549,
|
| 1562 |
+
"rewards/rejected": -0.5589307546615601,
|
| 1563 |
+
"step": 930
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.8612001832340815,
|
| 1567 |
+
"grad_norm": 0.9841431379318237,
|
| 1568 |
+
"learning_rate": 5.732924089870245e-08,
|
| 1569 |
+
"logits/chosen": -0.46664732694625854,
|
| 1570 |
+
"logits/rejected": -0.6183963418006897,
|
| 1571 |
+
"logps/chosen": -208.5711669921875,
|
| 1572 |
+
"logps/rejected": -259.8545837402344,
|
| 1573 |
+
"loss": 0.2725,
|
| 1574 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1575 |
+
"rewards/chosen": 1.87375009059906,
|
| 1576 |
+
"rewards/margins": 2.2145848274230957,
|
| 1577 |
+
"rewards/rejected": -0.34083452820777893,
|
| 1578 |
+
"step": 940
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.8703618873110398,
|
| 1582 |
+
"grad_norm": 1.0470377206802368,
|
| 1583 |
+
"learning_rate": 5.011275652893782e-08,
|
| 1584 |
+
"logits/chosen": -0.6211365461349487,
|
| 1585 |
+
"logits/rejected": -0.6796506643295288,
|
| 1586 |
+
"logps/chosen": -152.32177734375,
|
| 1587 |
+
"logps/rejected": -218.6947479248047,
|
| 1588 |
+
"loss": 0.2395,
|
| 1589 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1590 |
+
"rewards/chosen": 1.6267896890640259,
|
| 1591 |
+
"rewards/margins": 2.076831102371216,
|
| 1592 |
+
"rewards/rejected": -0.45004144310951233,
|
| 1593 |
+
"step": 950
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.8795235913879982,
|
| 1597 |
+
"grad_norm": 1.149141550064087,
|
| 1598 |
+
"learning_rate": 4.3357619788127634e-08,
|
| 1599 |
+
"logits/chosen": -0.6848306655883789,
|
| 1600 |
+
"logits/rejected": -0.7928074598312378,
|
| 1601 |
+
"logps/chosen": -200.9459228515625,
|
| 1602 |
+
"logps/rejected": -254.53952026367188,
|
| 1603 |
+
"loss": 0.2671,
|
| 1604 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1605 |
+
"rewards/chosen": 1.663435697555542,
|
| 1606 |
+
"rewards/margins": 1.9824724197387695,
|
| 1607 |
+
"rewards/rejected": -0.3190363943576813,
|
| 1608 |
+
"step": 960
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 0.8886852954649564,
|
| 1612 |
+
"grad_norm": 1.5327177047729492,
|
| 1613 |
+
"learning_rate": 3.707075789273306e-08,
|
| 1614 |
+
"logits/chosen": -0.6709850430488586,
|
| 1615 |
+
"logits/rejected": -0.6786261796951294,
|
| 1616 |
+
"logps/chosen": -146.47488403320312,
|
| 1617 |
+
"logps/rejected": -247.5612335205078,
|
| 1618 |
+
"loss": 0.2843,
|
| 1619 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1620 |
+
"rewards/chosen": 1.8335565328598022,
|
| 1621 |
+
"rewards/margins": 2.448019504547119,
|
| 1622 |
+
"rewards/rejected": -0.6144627332687378,
|
| 1623 |
+
"step": 970
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.8978469995419148,
|
| 1627 |
+
"grad_norm": 1.4160970449447632,
|
| 1628 |
+
"learning_rate": 3.125861785558015e-08,
|
| 1629 |
+
"logits/chosen": -0.7669690251350403,
|
| 1630 |
+
"logits/rejected": -0.8644639849662781,
|
| 1631 |
+
"logps/chosen": -169.50274658203125,
|
| 1632 |
+
"logps/rejected": -304.95098876953125,
|
| 1633 |
+
"loss": 0.2834,
|
| 1634 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1635 |
+
"rewards/chosen": 1.8083531856536865,
|
| 1636 |
+
"rewards/margins": 2.9228787422180176,
|
| 1637 |
+
"rewards/rejected": -1.114525556564331,
|
| 1638 |
+
"step": 980
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 0.9070087036188731,
|
| 1642 |
+
"grad_norm": 2.2089922428131104,
|
| 1643 |
+
"learning_rate": 2.592715987461702e-08,
|
| 1644 |
+
"logits/chosen": -0.7596947550773621,
|
| 1645 |
+
"logits/rejected": -0.7801377177238464,
|
| 1646 |
+
"logps/chosen": -217.3747100830078,
|
| 1647 |
+
"logps/rejected": -258.47796630859375,
|
| 1648 |
+
"loss": 0.3283,
|
| 1649 |
+
"rewards/accuracies": 0.875,
|
| 1650 |
+
"rewards/chosen": 1.5211076736450195,
|
| 1651 |
+
"rewards/margins": 1.6776663064956665,
|
| 1652 |
+
"rewards/rejected": -0.15655846893787384,
|
| 1653 |
+
"step": 990
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.9161704076958315,
|
| 1657 |
+
"grad_norm": 1.1835498809814453,
|
| 1658 |
+
"learning_rate": 2.108185122088546e-08,
|
| 1659 |
+
"logits/chosen": -0.7457928657531738,
|
| 1660 |
+
"logits/rejected": -0.6937334537506104,
|
| 1661 |
+
"logps/chosen": -169.40447998046875,
|
| 1662 |
+
"logps/rejected": -247.85061645507812,
|
| 1663 |
+
"loss": 0.268,
|
| 1664 |
+
"rewards/accuracies": 0.875,
|
| 1665 |
+
"rewards/chosen": 1.8612730503082275,
|
| 1666 |
+
"rewards/margins": 2.29799747467041,
|
| 1667 |
+
"rewards/rejected": -0.43672457337379456,
|
| 1668 |
+
"step": 1000
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.9161704076958315,
|
| 1672 |
+
"eval_logits/chosen": -0.6635109186172485,
|
| 1673 |
+
"eval_logits/rejected": -0.6690148711204529,
|
| 1674 |
+
"eval_logps/chosen": -200.81570434570312,
|
| 1675 |
+
"eval_logps/rejected": -276.8312072753906,
|
| 1676 |
+
"eval_loss": 0.29993191361427307,
|
| 1677 |
+
"eval_rewards/accuracies": 0.884393036365509,
|
| 1678 |
+
"eval_rewards/chosen": 1.6797596216201782,
|
| 1679 |
+
"eval_rewards/margins": 2.1726391315460205,
|
| 1680 |
+
"eval_rewards/rejected": -0.4928795397281647,
|
| 1681 |
+
"eval_runtime": 253.6729,
|
| 1682 |
+
"eval_samples_per_second": 10.892,
|
| 1683 |
+
"eval_steps_per_second": 1.364,
|
| 1684 |
+
"step": 1000
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.9253321117727897,
|
| 1688 |
+
"grad_norm": 1.9540088176727295,
|
| 1689 |
+
"learning_rate": 1.672766063197789e-08,
|
| 1690 |
+
"logits/chosen": -0.6848675608634949,
|
| 1691 |
+
"logits/rejected": -0.6749913692474365,
|
| 1692 |
+
"logps/chosen": -182.75892639160156,
|
| 1693 |
+
"logps/rejected": -234.47232055664062,
|
| 1694 |
+
"loss": 0.2621,
|
| 1695 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1696 |
+
"rewards/chosen": 1.729029655456543,
|
| 1697 |
+
"rewards/margins": 2.2015597820281982,
|
| 1698 |
+
"rewards/rejected": -0.47253018617630005,
|
| 1699 |
+
"step": 1010
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 0.934493815849748,
|
| 1703 |
+
"grad_norm": 1.5091657638549805,
|
| 1704 |
+
"learning_rate": 1.286905321672621e-08,
|
| 1705 |
+
"logits/chosen": -0.6353659629821777,
|
| 1706 |
+
"logits/rejected": -0.6381145715713501,
|
| 1707 |
+
"logps/chosen": -118.5170669555664,
|
| 1708 |
+
"logps/rejected": -205.03689575195312,
|
| 1709 |
+
"loss": 0.2602,
|
| 1710 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1711 |
+
"rewards/chosen": 1.9368598461151123,
|
| 1712 |
+
"rewards/margins": 2.6814284324645996,
|
| 1713 |
+
"rewards/rejected": -0.744568407535553,
|
| 1714 |
+
"step": 1020
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 0.9436555199267064,
|
| 1718 |
+
"grad_norm": 1.7073416709899902,
|
| 1719 |
+
"learning_rate": 9.509985876349491e-09,
|
| 1720 |
+
"logits/chosen": -0.61668461561203,
|
| 1721 |
+
"logits/rejected": -0.6262907385826111,
|
| 1722 |
+
"logps/chosen": -143.79531860351562,
|
| 1723 |
+
"logps/rejected": -229.5824737548828,
|
| 1724 |
+
"loss": 0.2564,
|
| 1725 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1726 |
+
"rewards/chosen": 1.7922176122665405,
|
| 1727 |
+
"rewards/margins": 2.186096429824829,
|
| 1728 |
+
"rewards/rejected": -0.3938787579536438,
|
| 1729 |
+
"step": 1030
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 0.9528172240036646,
|
| 1733 |
+
"grad_norm": 1.6124757528305054,
|
| 1734 |
+
"learning_rate": 6.6539032467546885e-09,
|
| 1735 |
+
"logits/chosen": -0.790496289730072,
|
| 1736 |
+
"logits/rejected": -0.7677043080329895,
|
| 1737 |
+
"logps/chosen": -256.6092834472656,
|
| 1738 |
+
"logps/rejected": -325.1084289550781,
|
| 1739 |
+
"loss": 0.262,
|
| 1740 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1741 |
+
"rewards/chosen": 1.504821538925171,
|
| 1742 |
+
"rewards/margins": 2.0350840091705322,
|
| 1743 |
+
"rewards/rejected": -0.5302623510360718,
|
| 1744 |
+
"step": 1040
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 0.961978928080623,
|
| 1748 |
+
"grad_norm": 1.0290203094482422,
|
| 1749 |
+
"learning_rate": 4.303734166152706e-09,
|
| 1750 |
+
"logits/chosen": -0.7079821825027466,
|
| 1751 |
+
"logits/rejected": -0.6711887121200562,
|
| 1752 |
+
"logps/chosen": -177.19007873535156,
|
| 1753 |
+
"logps/rejected": -210.00613403320312,
|
| 1754 |
+
"loss": 0.3015,
|
| 1755 |
+
"rewards/accuracies": 0.875,
|
| 1756 |
+
"rewards/chosen": 1.794146180152893,
|
| 1757 |
+
"rewards/margins": 2.358898639678955,
|
| 1758 |
+
"rewards/rejected": -0.564752459526062,
|
| 1759 |
+
"step": 1050
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.9711406321575813,
|
| 1763 |
+
"grad_norm": 1.0389478206634521,
|
| 1764 |
+
"learning_rate": 2.4618886716110676e-09,
|
| 1765 |
+
"logits/chosen": -0.6768301725387573,
|
| 1766 |
+
"logits/rejected": -0.6700726747512817,
|
| 1767 |
+
"logps/chosen": -167.974609375,
|
| 1768 |
+
"logps/rejected": -247.9666290283203,
|
| 1769 |
+
"loss": 0.2928,
|
| 1770 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1771 |
+
"rewards/chosen": 1.5335534811019897,
|
| 1772 |
+
"rewards/margins": 1.9949595928192139,
|
| 1773 |
+
"rewards/rejected": -0.4614059329032898,
|
| 1774 |
+
"step": 1060
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.9803023362345397,
|
| 1778 |
+
"grad_norm": 0.8039052486419678,
|
| 1779 |
+
"learning_rate": 1.1302555276238579e-09,
|
| 1780 |
+
"logits/chosen": -0.6699908375740051,
|
| 1781 |
+
"logits/rejected": -0.7130982279777527,
|
| 1782 |
+
"logps/chosen": -109.6313247680664,
|
| 1783 |
+
"logps/rejected": -177.20127868652344,
|
| 1784 |
+
"loss": 0.232,
|
| 1785 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1786 |
+
"rewards/chosen": 1.9235477447509766,
|
| 1787 |
+
"rewards/margins": 2.2871975898742676,
|
| 1788 |
+
"rewards/rejected": -0.3636501729488373,
|
| 1789 |
+
"step": 1070
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 0.9894640403114979,
|
| 1793 |
+
"grad_norm": 2.4051105976104736,
|
| 1794 |
+
"learning_rate": 3.102002892329536e-10,
|
| 1795 |
+
"logits/chosen": -0.6233320832252502,
|
| 1796 |
+
"logits/rejected": -0.6594210267066956,
|
| 1797 |
+
"logps/chosen": -142.9510040283203,
|
| 1798 |
+
"logps/rejected": -230.3099365234375,
|
| 1799 |
+
"loss": 0.2467,
|
| 1800 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1801 |
+
"rewards/chosen": 1.8195747137069702,
|
| 1802 |
+
"rewards/margins": 2.5644664764404297,
|
| 1803 |
+
"rewards/rejected": -0.7448917031288147,
|
| 1804 |
+
"step": 1080
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 0.9986257443884563,
|
| 1808 |
+
"grad_norm": 1.9035027027130127,
|
| 1809 |
+
"learning_rate": 2.5639016871248366e-12,
|
| 1810 |
+
"logits/chosen": -0.608766496181488,
|
| 1811 |
+
"logits/rejected": -0.5473885536193848,
|
| 1812 |
+
"logps/chosen": -232.18392944335938,
|
| 1813 |
+
"logps/rejected": -249.3759002685547,
|
| 1814 |
+
"loss": 0.2929,
|
| 1815 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1816 |
+
"rewards/chosen": 1.7724415063858032,
|
| 1817 |
+
"rewards/margins": 2.1429972648620605,
|
| 1818 |
+
"rewards/rejected": -0.3705558478832245,
|
| 1819 |
+
"step": 1090
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 0.9995419147961521,
|
| 1823 |
+
"step": 1091,
|
| 1824 |
+
"total_flos": 0.0,
|
| 1825 |
+
"train_loss": 0.37405444834616947,
|
| 1826 |
+
"train_runtime": 8738.9435,
|
| 1827 |
+
"train_samples_per_second": 3.996,
|
| 1828 |
+
"train_steps_per_second": 0.125
|
| 1829 |
+
}
|
| 1830 |
+
],
|
| 1831 |
+
"logging_steps": 10,
|
| 1832 |
+
"max_steps": 1091,
|
| 1833 |
+
"num_input_tokens_seen": 0,
|
| 1834 |
+
"num_train_epochs": 1,
|
| 1835 |
+
"save_steps": 100,
|
| 1836 |
+
"stateful_callbacks": {
|
| 1837 |
+
"TrainerControl": {
|
| 1838 |
+
"args": {
|
| 1839 |
+
"should_epoch_stop": false,
|
| 1840 |
+
"should_evaluate": false,
|
| 1841 |
+
"should_log": false,
|
| 1842 |
+
"should_save": true,
|
| 1843 |
+
"should_training_stop": true
|
| 1844 |
+
},
|
| 1845 |
+
"attributes": {}
|
| 1846 |
+
}
|
| 1847 |
+
},
|
| 1848 |
+
"total_flos": 0.0,
|
| 1849 |
+
"train_batch_size": 1,
|
| 1850 |
+
"trial_name": null,
|
| 1851 |
+
"trial_params": null
|
| 1852 |
+
}
|