Model save
Browse files
README.md
ADDED
|
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 3 |
+
library_name: peft
|
| 4 |
+
license: apache-2.0
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-dpo-qlora-uf6k-5e-7
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-dpo-qlora-uf6k-5e-7
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6889
|
| 22 |
+
- Rewards/chosen: 0.0031
|
| 23 |
+
- Rewards/rejected: -0.0066
|
| 24 |
+
- Rewards/accuracies: 0.6760
|
| 25 |
+
- Rewards/margins: 0.0098
|
| 26 |
+
- Rewards/margins Max: 0.0449
|
| 27 |
+
- Rewards/margins Min: -0.0216
|
| 28 |
+
- Rewards/margins Std: 0.0219
|
| 29 |
+
- Logps/rejected: -259.2438
|
| 30 |
+
- Logps/chosen: -284.2815
|
| 31 |
+
- Logits/rejected: -2.7640
|
| 32 |
+
- Logits/chosen: -2.8026
|
| 33 |
+
|
| 34 |
+
## Model description
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Intended uses & limitations
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training and evaluation data
|
| 43 |
+
|
| 44 |
+
More information needed
|
| 45 |
+
|
| 46 |
+
## Training procedure
|
| 47 |
+
|
| 48 |
+
### Training hyperparameters
|
| 49 |
+
|
| 50 |
+
The following hyperparameters were used during training:
|
| 51 |
+
- learning_rate: 5e-07
|
| 52 |
+
- train_batch_size: 4
|
| 53 |
+
- eval_batch_size: 8
|
| 54 |
+
- seed: 42
|
| 55 |
+
- distributed_type: multi-GPU
|
| 56 |
+
- num_devices: 2
|
| 57 |
+
- gradient_accumulation_steps: 2
|
| 58 |
+
- total_train_batch_size: 16
|
| 59 |
+
- total_eval_batch_size: 16
|
| 60 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 61 |
+
- lr_scheduler_type: cosine
|
| 62 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 63 |
+
- num_epochs: 1
|
| 64 |
+
|
| 65 |
+
### Training results
|
| 66 |
+
|
| 67 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Rewards/margins Max | Rewards/margins Min | Rewards/margins Std | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 68 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:-------------------:|:-------------------:|:-------------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 69 |
+
| 0.6914 | 0.3 | 100 | 0.6915 | -0.0004 | -0.0041 | 0.6460 | 0.0037 | 0.0187 | -0.0095 | 0.0093 | -258.9901 | -284.6319 | -2.7677 | -2.8065 |
|
| 70 |
+
| 0.6884 | 0.61 | 200 | 0.6895 | 0.0023 | -0.0061 | 0.6850 | 0.0084 | 0.0389 | -0.0189 | 0.0190 | -259.1880 | -284.3611 | -2.7665 | -2.8049 |
|
| 71 |
+
| 0.6873 | 0.91 | 300 | 0.6889 | 0.0031 | -0.0066 | 0.6760 | 0.0098 | 0.0449 | -0.0216 | 0.0219 | -259.2438 | -284.2815 | -2.7640 | -2.8026 |
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
### Framework versions
|
| 75 |
+
|
| 76 |
+
- PEFT 0.7.1
|
| 77 |
+
- Transformers 4.39.0.dev0
|
| 78 |
+
- Pytorch 2.1.2+cu121
|
| 79 |
+
- Datasets 2.14.6
|
| 80 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671150064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af91c8567e56a7a7d7a09a5beac038fd988d31153db35e7f94e82025afe71566
|
| 3 |
size 671150064
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6900796745323483,
|
| 4 |
+
"train_runtime": 3893.3874,
|
| 5 |
+
"train_samples": 5263,
|
| 6 |
+
"train_samples_per_second": 1.352,
|
| 7 |
+
"train_steps_per_second": 0.085
|
| 8 |
+
}
|
runs/Jul25_08-24-43_notebook-deployment-48-7d9b6c99-khd85/events.out.tfevents.1721895976.notebook-deployment-48-7d9b6c99-khd85.3153936.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f64095a2cf0a5cbe4b46c1acb4cc7651c7e7c4c6308eb27d95cd8f0f6a50f253
|
| 3 |
+
size 37307
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6900796745323483,
|
| 4 |
+
"train_runtime": 3893.3874,
|
| 5 |
+
"train_samples": 5263,
|
| 6 |
+
"train_samples_per_second": 1.352,
|
| 7 |
+
"train_steps_per_second": 0.085
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,681 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 329,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 2.0906090021590473,
|
| 14 |
+
"learning_rate": 1.5151515151515152e-08,
|
| 15 |
+
"logits/chosen": -2.6820077896118164,
|
| 16 |
+
"logits/rejected": -2.6930205821990967,
|
| 17 |
+
"logps/chosen": -281.2528381347656,
|
| 18 |
+
"logps/rejected": -258.0622253417969,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/margins_max": 0.0,
|
| 24 |
+
"rewards/margins_min": 0.0,
|
| 25 |
+
"rewards/margins_std": 0.0,
|
| 26 |
+
"rewards/rejected": 0.0,
|
| 27 |
+
"step": 1
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.03,
|
| 31 |
+
"grad_norm": 2.1561337868153565,
|
| 32 |
+
"learning_rate": 1.5151515151515152e-07,
|
| 33 |
+
"logits/chosen": -2.7683067321777344,
|
| 34 |
+
"logits/rejected": -2.7538461685180664,
|
| 35 |
+
"logps/chosen": -284.59912109375,
|
| 36 |
+
"logps/rejected": -249.83580017089844,
|
| 37 |
+
"loss": 0.6931,
|
| 38 |
+
"rewards/accuracies": 0.3888888955116272,
|
| 39 |
+
"rewards/chosen": 5.0317983550485224e-05,
|
| 40 |
+
"rewards/margins": -0.00015015894314274192,
|
| 41 |
+
"rewards/margins_max": 0.0020335663575679064,
|
| 42 |
+
"rewards/margins_min": -0.0025187418796122074,
|
| 43 |
+
"rewards/margins_std": 0.0020784963853657246,
|
| 44 |
+
"rewards/rejected": 0.00020047693396918476,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.06,
|
| 49 |
+
"grad_norm": 1.9751920510122447,
|
| 50 |
+
"learning_rate": 3.0303030303030305e-07,
|
| 51 |
+
"logits/chosen": -2.8347439765930176,
|
| 52 |
+
"logits/rejected": -2.7819018363952637,
|
| 53 |
+
"logps/chosen": -291.50921630859375,
|
| 54 |
+
"logps/rejected": -270.4449768066406,
|
| 55 |
+
"loss": 0.693,
|
| 56 |
+
"rewards/accuracies": 0.5625,
|
| 57 |
+
"rewards/chosen": 0.00022530628484673798,
|
| 58 |
+
"rewards/margins": 0.0007139825029298663,
|
| 59 |
+
"rewards/margins_max": 0.004345210734754801,
|
| 60 |
+
"rewards/margins_min": -0.002943573985248804,
|
| 61 |
+
"rewards/margins_std": 0.0032876902259886265,
|
| 62 |
+
"rewards/rejected": -0.0004886762471869588,
|
| 63 |
+
"step": 20
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.09,
|
| 67 |
+
"grad_norm": 1.6654288543237405,
|
| 68 |
+
"learning_rate": 4.545454545454545e-07,
|
| 69 |
+
"logits/chosen": -2.8627753257751465,
|
| 70 |
+
"logits/rejected": -2.8151745796203613,
|
| 71 |
+
"logps/chosen": -259.2825927734375,
|
| 72 |
+
"logps/rejected": -227.37350463867188,
|
| 73 |
+
"loss": 0.6932,
|
| 74 |
+
"rewards/accuracies": 0.5,
|
| 75 |
+
"rewards/chosen": -0.00037297833478078246,
|
| 76 |
+
"rewards/margins": -0.00020202626183163375,
|
| 77 |
+
"rewards/margins_max": 0.0030375297646969557,
|
| 78 |
+
"rewards/margins_min": -0.0033374775666743517,
|
| 79 |
+
"rewards/margins_std": 0.0028792533557862043,
|
| 80 |
+
"rewards/rejected": -0.00017095205839723349,
|
| 81 |
+
"step": 30
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.12,
|
| 85 |
+
"grad_norm": 1.6057052994928989,
|
| 86 |
+
"learning_rate": 4.993103596812268e-07,
|
| 87 |
+
"logits/chosen": -2.8291430473327637,
|
| 88 |
+
"logits/rejected": -2.7638001441955566,
|
| 89 |
+
"logps/chosen": -317.513916015625,
|
| 90 |
+
"logps/rejected": -224.7698211669922,
|
| 91 |
+
"loss": 0.6927,
|
| 92 |
+
"rewards/accuracies": 0.625,
|
| 93 |
+
"rewards/chosen": 0.000282805209280923,
|
| 94 |
+
"rewards/margins": 0.0011336280731484294,
|
| 95 |
+
"rewards/margins_max": 0.005270844791084528,
|
| 96 |
+
"rewards/margins_min": -0.002321633044630289,
|
| 97 |
+
"rewards/margins_std": 0.003387246746569872,
|
| 98 |
+
"rewards/rejected": -0.000850822776556015,
|
| 99 |
+
"step": 40
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.15,
|
| 103 |
+
"grad_norm": 1.75475734119915,
|
| 104 |
+
"learning_rate": 4.959416858332709e-07,
|
| 105 |
+
"logits/chosen": -2.79063081741333,
|
| 106 |
+
"logits/rejected": -2.804368495941162,
|
| 107 |
+
"logps/chosen": -242.9667510986328,
|
| 108 |
+
"logps/rejected": -280.0011901855469,
|
| 109 |
+
"loss": 0.6926,
|
| 110 |
+
"rewards/accuracies": 0.574999988079071,
|
| 111 |
+
"rewards/chosen": -0.00031891773687675595,
|
| 112 |
+
"rewards/margins": 0.0008532041683793068,
|
| 113 |
+
"rewards/margins_max": 0.004698004573583603,
|
| 114 |
+
"rewards/margins_min": -0.002724443329498172,
|
| 115 |
+
"rewards/margins_std": 0.00329922279343009,
|
| 116 |
+
"rewards/rejected": -0.0011721218470484018,
|
| 117 |
+
"step": 50
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.18,
|
| 121 |
+
"grad_norm": 1.91854731579599,
|
| 122 |
+
"learning_rate": 4.898051734555674e-07,
|
| 123 |
+
"logits/chosen": -2.8335373401641846,
|
| 124 |
+
"logits/rejected": -2.8440303802490234,
|
| 125 |
+
"logps/chosen": -321.90625,
|
| 126 |
+
"logps/rejected": -283.37994384765625,
|
| 127 |
+
"loss": 0.6921,
|
| 128 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 129 |
+
"rewards/chosen": 0.00014833270688541234,
|
| 130 |
+
"rewards/margins": 0.0021868678741157055,
|
| 131 |
+
"rewards/margins_max": 0.008168894797563553,
|
| 132 |
+
"rewards/margins_min": -0.0031033740378916264,
|
| 133 |
+
"rewards/margins_std": 0.005018442869186401,
|
| 134 |
+
"rewards/rejected": -0.0020385351963341236,
|
| 135 |
+
"step": 60
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.21,
|
| 139 |
+
"grad_norm": 1.5964843630528198,
|
| 140 |
+
"learning_rate": 4.809698831278217e-07,
|
| 141 |
+
"logits/chosen": -2.748741865158081,
|
| 142 |
+
"logits/rejected": -2.735199213027954,
|
| 143 |
+
"logps/chosen": -266.52606201171875,
|
| 144 |
+
"logps/rejected": -246.6175079345703,
|
| 145 |
+
"loss": 0.6922,
|
| 146 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 147 |
+
"rewards/chosen": -0.0004858696775045246,
|
| 148 |
+
"rewards/margins": 0.0019440820906311274,
|
| 149 |
+
"rewards/margins_max": 0.008044283837080002,
|
| 150 |
+
"rewards/margins_min": -0.003816543845459819,
|
| 151 |
+
"rewards/margins_std": 0.005177702754735947,
|
| 152 |
+
"rewards/rejected": -0.002429951447993517,
|
| 153 |
+
"step": 70
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.24,
|
| 157 |
+
"grad_norm": 2.1014040068240663,
|
| 158 |
+
"learning_rate": 4.6953524759527053e-07,
|
| 159 |
+
"logits/chosen": -2.8426356315612793,
|
| 160 |
+
"logits/rejected": -2.8158562183380127,
|
| 161 |
+
"logps/chosen": -282.353515625,
|
| 162 |
+
"logps/rejected": -275.220458984375,
|
| 163 |
+
"loss": 0.6918,
|
| 164 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 165 |
+
"rewards/chosen": -0.000800526118837297,
|
| 166 |
+
"rewards/margins": 0.0022392510436475277,
|
| 167 |
+
"rewards/margins_max": 0.00997895933687687,
|
| 168 |
+
"rewards/margins_min": -0.005192113574594259,
|
| 169 |
+
"rewards/margins_std": 0.0067059798166155815,
|
| 170 |
+
"rewards/rejected": -0.0030397772789001465,
|
| 171 |
+
"step": 80
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.27,
|
| 175 |
+
"grad_norm": 1.9914287244112958,
|
| 176 |
+
"learning_rate": 4.5562995274820283e-07,
|
| 177 |
+
"logits/chosen": -2.7992029190063477,
|
| 178 |
+
"logits/rejected": -2.746138095855713,
|
| 179 |
+
"logps/chosen": -295.78399658203125,
|
| 180 |
+
"logps/rejected": -291.9333190917969,
|
| 181 |
+
"loss": 0.6919,
|
| 182 |
+
"rewards/accuracies": 0.5625,
|
| 183 |
+
"rewards/chosen": -0.002320217899978161,
|
| 184 |
+
"rewards/margins": 0.001351921702735126,
|
| 185 |
+
"rewards/margins_max": 0.010480575263500214,
|
| 186 |
+
"rewards/margins_min": -0.009322223253548145,
|
| 187 |
+
"rewards/margins_std": 0.008863108232617378,
|
| 188 |
+
"rewards/rejected": -0.0036721397191286087,
|
| 189 |
+
"step": 90
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.3,
|
| 193 |
+
"grad_norm": 1.6705180570693485,
|
| 194 |
+
"learning_rate": 4.394104893853007e-07,
|
| 195 |
+
"logits/chosen": -2.896794557571411,
|
| 196 |
+
"logits/rejected": -2.85756254196167,
|
| 197 |
+
"logps/chosen": -273.5906982421875,
|
| 198 |
+
"logps/rejected": -257.73284912109375,
|
| 199 |
+
"loss": 0.6914,
|
| 200 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 201 |
+
"rewards/chosen": -0.0012845676392316818,
|
| 202 |
+
"rewards/margins": 0.005008908919990063,
|
| 203 |
+
"rewards/margins_max": 0.013928805477917194,
|
| 204 |
+
"rewards/margins_min": -0.003106380347162485,
|
| 205 |
+
"rewards/margins_std": 0.007625125348567963,
|
| 206 |
+
"rewards/rejected": -0.006293477024883032,
|
| 207 |
+
"step": 100
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.3,
|
| 211 |
+
"eval_logits/chosen": -2.806475877761841,
|
| 212 |
+
"eval_logits/rejected": -2.767702102661133,
|
| 213 |
+
"eval_logps/chosen": -284.6319274902344,
|
| 214 |
+
"eval_logps/rejected": -258.9901123046875,
|
| 215 |
+
"eval_loss": 0.691525399684906,
|
| 216 |
+
"eval_rewards/accuracies": 0.6460000276565552,
|
| 217 |
+
"eval_rewards/chosen": -0.00038527295691892505,
|
| 218 |
+
"eval_rewards/margins": 0.003726556431502104,
|
| 219 |
+
"eval_rewards/margins_max": 0.01873905211687088,
|
| 220 |
+
"eval_rewards/margins_min": -0.009459242224693298,
|
| 221 |
+
"eval_rewards/margins_std": 0.00931489747017622,
|
| 222 |
+
"eval_rewards/rejected": -0.004111829213798046,
|
| 223 |
+
"eval_runtime": 428.4684,
|
| 224 |
+
"eval_samples_per_second": 4.668,
|
| 225 |
+
"eval_steps_per_second": 0.292,
|
| 226 |
+
"step": 100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.33,
|
| 230 |
+
"grad_norm": 2.1453641236519685,
|
| 231 |
+
"learning_rate": 4.2105939205932005e-07,
|
| 232 |
+
"logits/chosen": -2.7631096839904785,
|
| 233 |
+
"logits/rejected": -2.746663808822632,
|
| 234 |
+
"logps/chosen": -311.8393249511719,
|
| 235 |
+
"logps/rejected": -235.84280395507812,
|
| 236 |
+
"loss": 0.6911,
|
| 237 |
+
"rewards/accuracies": 0.625,
|
| 238 |
+
"rewards/chosen": -0.0008311712299473584,
|
| 239 |
+
"rewards/margins": 0.0033794320188462734,
|
| 240 |
+
"rewards/margins_max": 0.013278109021484852,
|
| 241 |
+
"rewards/margins_min": -0.00541637372225523,
|
| 242 |
+
"rewards/margins_std": 0.008299448527395725,
|
| 243 |
+
"rewards/rejected": -0.0042106034234166145,
|
| 244 |
+
"step": 110
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.36,
|
| 248 |
+
"grad_norm": 2.024896986425123,
|
| 249 |
+
"learning_rate": 4.0078318482522114e-07,
|
| 250 |
+
"logits/chosen": -2.7521708011627197,
|
| 251 |
+
"logits/rejected": -2.750868082046509,
|
| 252 |
+
"logps/chosen": -323.51666259765625,
|
| 253 |
+
"logps/rejected": -274.75970458984375,
|
| 254 |
+
"loss": 0.6909,
|
| 255 |
+
"rewards/accuracies": 0.75,
|
| 256 |
+
"rewards/chosen": 0.0004785500350408256,
|
| 257 |
+
"rewards/margins": 0.004080395679920912,
|
| 258 |
+
"rewards/margins_max": 0.015328818932175636,
|
| 259 |
+
"rewards/margins_min": -0.0073760440573096275,
|
| 260 |
+
"rewards/margins_std": 0.00990099273622036,
|
| 261 |
+
"rewards/rejected": -0.0036018460523337126,
|
| 262 |
+
"step": 120
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.4,
|
| 266 |
+
"grad_norm": 1.6346525930252072,
|
| 267 |
+
"learning_rate": 3.7881005700938627e-07,
|
| 268 |
+
"logits/chosen": -2.8206729888916016,
|
| 269 |
+
"logits/rejected": -2.8308663368225098,
|
| 270 |
+
"logps/chosen": -266.37469482421875,
|
| 271 |
+
"logps/rejected": -234.52035522460938,
|
| 272 |
+
"loss": 0.6906,
|
| 273 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 274 |
+
"rewards/chosen": 0.00018907712365034968,
|
| 275 |
+
"rewards/margins": 0.00421659741550684,
|
| 276 |
+
"rewards/margins_max": 0.015676384791731834,
|
| 277 |
+
"rewards/margins_min": -0.007556927390396595,
|
| 278 |
+
"rewards/margins_std": 0.010246575810015202,
|
| 279 |
+
"rewards/rejected": -0.004027520306408405,
|
| 280 |
+
"step": 130
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.43,
|
| 284 |
+
"grad_norm": 1.9044203185149946,
|
| 285 |
+
"learning_rate": 3.5538729515692354e-07,
|
| 286 |
+
"logits/chosen": -2.780360460281372,
|
| 287 |
+
"logits/rejected": -2.7639012336730957,
|
| 288 |
+
"logps/chosen": -294.11309814453125,
|
| 289 |
+
"logps/rejected": -270.84710693359375,
|
| 290 |
+
"loss": 0.6896,
|
| 291 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 292 |
+
"rewards/chosen": 0.0028018890880048275,
|
| 293 |
+
"rewards/margins": 0.007480897009372711,
|
| 294 |
+
"rewards/margins_max": 0.021374408155679703,
|
| 295 |
+
"rewards/margins_min": -0.0061719887889921665,
|
| 296 |
+
"rewards/margins_std": 0.01222699973732233,
|
| 297 |
+
"rewards/rejected": -0.004679008387029171,
|
| 298 |
+
"step": 140
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.46,
|
| 302 |
+
"grad_norm": 1.4256559970133287,
|
| 303 |
+
"learning_rate": 3.3077850005803125e-07,
|
| 304 |
+
"logits/chosen": -2.8410263061523438,
|
| 305 |
+
"logits/rejected": -2.8195314407348633,
|
| 306 |
+
"logps/chosen": -270.49615478515625,
|
| 307 |
+
"logps/rejected": -245.65200805664062,
|
| 308 |
+
"loss": 0.6903,
|
| 309 |
+
"rewards/accuracies": 0.699999988079071,
|
| 310 |
+
"rewards/chosen": 0.0014524383004754782,
|
| 311 |
+
"rewards/margins": 0.006768654100596905,
|
| 312 |
+
"rewards/margins_max": 0.025039460510015488,
|
| 313 |
+
"rewards/margins_min": -0.01076546311378479,
|
| 314 |
+
"rewards/margins_std": 0.015858832746744156,
|
| 315 |
+
"rewards/rejected": -0.0053162164986133575,
|
| 316 |
+
"step": 150
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.49,
|
| 320 |
+
"grad_norm": 2.1265057109843077,
|
| 321 |
+
"learning_rate": 3.0526062017313247e-07,
|
| 322 |
+
"logits/chosen": -2.79884672164917,
|
| 323 |
+
"logits/rejected": -2.7815585136413574,
|
| 324 |
+
"logps/chosen": -255.3964080810547,
|
| 325 |
+
"logps/rejected": -241.00271606445312,
|
| 326 |
+
"loss": 0.6909,
|
| 327 |
+
"rewards/accuracies": 0.637499988079071,
|
| 328 |
+
"rewards/chosen": 0.0010771710658445954,
|
| 329 |
+
"rewards/margins": 0.005134746432304382,
|
| 330 |
+
"rewards/margins_max": 0.022996146231889725,
|
| 331 |
+
"rewards/margins_min": -0.009730304591357708,
|
| 332 |
+
"rewards/margins_std": 0.014861812815070152,
|
| 333 |
+
"rewards/rejected": -0.004057575948536396,
|
| 334 |
+
"step": 160
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.52,
|
| 338 |
+
"grad_norm": 1.59020242230242,
|
| 339 |
+
"learning_rate": 2.791208348427426e-07,
|
| 340 |
+
"logits/chosen": -2.814671039581299,
|
| 341 |
+
"logits/rejected": -2.732504367828369,
|
| 342 |
+
"logps/chosen": -303.4354553222656,
|
| 343 |
+
"logps/rejected": -273.4683837890625,
|
| 344 |
+
"loss": 0.6887,
|
| 345 |
+
"rewards/accuracies": 0.699999988079071,
|
| 346 |
+
"rewards/chosen": 0.002831272780895233,
|
| 347 |
+
"rewards/margins": 0.008308259770274162,
|
| 348 |
+
"rewards/margins_max": 0.02398056350648403,
|
| 349 |
+
"rewards/margins_min": -0.007372391410171986,
|
| 350 |
+
"rewards/margins_std": 0.014078010804951191,
|
| 351 |
+
"rewards/rejected": -0.005476987920701504,
|
| 352 |
+
"step": 170
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.55,
|
| 356 |
+
"grad_norm": 1.791424187804565,
|
| 357 |
+
"learning_rate": 2.526533223585641e-07,
|
| 358 |
+
"logits/chosen": -2.8398988246917725,
|
| 359 |
+
"logits/rejected": -2.775310754776001,
|
| 360 |
+
"logps/chosen": -256.0347595214844,
|
| 361 |
+
"logps/rejected": -229.332763671875,
|
| 362 |
+
"loss": 0.6897,
|
| 363 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 364 |
+
"rewards/chosen": 0.0009899451397359371,
|
| 365 |
+
"rewards/margins": 0.005457731895148754,
|
| 366 |
+
"rewards/margins_max": 0.021505217999219894,
|
| 367 |
+
"rewards/margins_min": -0.008436702191829681,
|
| 368 |
+
"rewards/margins_std": 0.013385000638663769,
|
| 369 |
+
"rewards/rejected": -0.00446778628975153,
|
| 370 |
+
"step": 180
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.58,
|
| 374 |
+
"grad_norm": 1.7305988753672774,
|
| 375 |
+
"learning_rate": 2.261559492680755e-07,
|
| 376 |
+
"logits/chosen": -2.781790256500244,
|
| 377 |
+
"logits/rejected": -2.7643322944641113,
|
| 378 |
+
"logps/chosen": -300.09393310546875,
|
| 379 |
+
"logps/rejected": -271.13116455078125,
|
| 380 |
+
"loss": 0.6891,
|
| 381 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 382 |
+
"rewards/chosen": 0.004870180506259203,
|
| 383 |
+
"rewards/margins": 0.0101470947265625,
|
| 384 |
+
"rewards/margins_max": 0.03561341017484665,
|
| 385 |
+
"rewards/margins_min": -0.00919102318584919,
|
| 386 |
+
"rewards/margins_std": 0.019990354776382446,
|
| 387 |
+
"rewards/rejected": -0.005276912357658148,
|
| 388 |
+
"step": 190
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.61,
|
| 392 |
+
"grad_norm": 2.169958133205736,
|
| 393 |
+
"learning_rate": 1.9992691817133024e-07,
|
| 394 |
+
"logits/chosen": -2.7859396934509277,
|
| 395 |
+
"logits/rejected": -2.755178213119507,
|
| 396 |
+
"logps/chosen": -281.18170166015625,
|
| 397 |
+
"logps/rejected": -288.84930419921875,
|
| 398 |
+
"loss": 0.6884,
|
| 399 |
+
"rewards/accuracies": 0.699999988079071,
|
| 400 |
+
"rewards/chosen": 0.0041890377178788185,
|
| 401 |
+
"rewards/margins": 0.009892629459500313,
|
| 402 |
+
"rewards/margins_max": 0.03310906141996384,
|
| 403 |
+
"rewards/margins_min": -0.012689237482845783,
|
| 404 |
+
"rewards/margins_std": 0.02011021040380001,
|
| 405 |
+
"rewards/rejected": -0.005703592207282782,
|
| 406 |
+
"step": 200
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.61,
|
| 410 |
+
"eval_logits/chosen": -2.804927349090576,
|
| 411 |
+
"eval_logits/rejected": -2.766470432281494,
|
| 412 |
+
"eval_logps/chosen": -284.3610534667969,
|
| 413 |
+
"eval_logps/rejected": -259.1879577636719,
|
| 414 |
+
"eval_loss": 0.6895014643669128,
|
| 415 |
+
"eval_rewards/accuracies": 0.6850000023841858,
|
| 416 |
+
"eval_rewards/chosen": 0.0023234861437231302,
|
| 417 |
+
"eval_rewards/margins": 0.008414038456976414,
|
| 418 |
+
"eval_rewards/margins_max": 0.038945525884628296,
|
| 419 |
+
"eval_rewards/margins_min": -0.018883490934967995,
|
| 420 |
+
"eval_rewards/margins_std": 0.018971558660268784,
|
| 421 |
+
"eval_rewards/rejected": -0.006090551149100065,
|
| 422 |
+
"eval_runtime": 427.7798,
|
| 423 |
+
"eval_samples_per_second": 4.675,
|
| 424 |
+
"eval_steps_per_second": 0.292,
|
| 425 |
+
"step": 200
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.64,
|
| 429 |
+
"grad_norm": 1.9906194761669704,
|
| 430 |
+
"learning_rate": 1.742614117358029e-07,
|
| 431 |
+
"logits/chosen": -2.80131196975708,
|
| 432 |
+
"logits/rejected": -2.7576537132263184,
|
| 433 |
+
"logps/chosen": -304.849853515625,
|
| 434 |
+
"logps/rejected": -289.08197021484375,
|
| 435 |
+
"loss": 0.6877,
|
| 436 |
+
"rewards/accuracies": 0.699999988079071,
|
| 437 |
+
"rewards/chosen": 0.0046735843643546104,
|
| 438 |
+
"rewards/margins": 0.012557747773826122,
|
| 439 |
+
"rewards/margins_max": 0.03481978923082352,
|
| 440 |
+
"rewards/margins_min": -0.00802917592227459,
|
| 441 |
+
"rewards/margins_std": 0.019201457500457764,
|
| 442 |
+
"rewards/rejected": -0.007884165272116661,
|
| 443 |
+
"step": 210
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.67,
|
| 447 |
+
"grad_norm": 1.9658311065665528,
|
| 448 |
+
"learning_rate": 1.4944827069769122e-07,
|
| 449 |
+
"logits/chosen": -2.851292133331299,
|
| 450 |
+
"logits/rejected": -2.8257217407226562,
|
| 451 |
+
"logps/chosen": -312.4863586425781,
|
| 452 |
+
"logps/rejected": -266.73626708984375,
|
| 453 |
+
"loss": 0.6891,
|
| 454 |
+
"rewards/accuracies": 0.675000011920929,
|
| 455 |
+
"rewards/chosen": 0.004867873154580593,
|
| 456 |
+
"rewards/margins": 0.008174732327461243,
|
| 457 |
+
"rewards/margins_max": 0.028449540957808495,
|
| 458 |
+
"rewards/margins_min": -0.011079356074333191,
|
| 459 |
+
"rewards/margins_std": 0.01735488697886467,
|
| 460 |
+
"rewards/rejected": -0.003306858241558075,
|
| 461 |
+
"step": 220
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.7,
|
| 465 |
+
"grad_norm": 1.8987994692738805,
|
| 466 |
+
"learning_rate": 1.2576674323558928e-07,
|
| 467 |
+
"logits/chosen": -2.821254014968872,
|
| 468 |
+
"logits/rejected": -2.8421223163604736,
|
| 469 |
+
"logps/chosen": -288.6875,
|
| 470 |
+
"logps/rejected": -263.0277099609375,
|
| 471 |
+
"loss": 0.6906,
|
| 472 |
+
"rewards/accuracies": 0.574999988079071,
|
| 473 |
+
"rewards/chosen": -0.00041991579928435385,
|
| 474 |
+
"rewards/margins": 0.0022654212079942226,
|
| 475 |
+
"rewards/margins_max": 0.024668732658028603,
|
| 476 |
+
"rewards/margins_min": -0.022191215306520462,
|
| 477 |
+
"rewards/margins_std": 0.020731808617711067,
|
| 478 |
+
"rewards/rejected": -0.002685337094590068,
|
| 479 |
+
"step": 230
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.73,
|
| 483 |
+
"grad_norm": 2.049682090113544,
|
| 484 |
+
"learning_rate": 1.0348334229922676e-07,
|
| 485 |
+
"logits/chosen": -2.877260684967041,
|
| 486 |
+
"logits/rejected": -2.8300554752349854,
|
| 487 |
+
"logps/chosen": -290.80633544921875,
|
| 488 |
+
"logps/rejected": -275.846435546875,
|
| 489 |
+
"loss": 0.6893,
|
| 490 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 491 |
+
"rewards/chosen": 0.0021143355406820774,
|
| 492 |
+
"rewards/margins": 0.00877899769693613,
|
| 493 |
+
"rewards/margins_max": 0.03138250857591629,
|
| 494 |
+
"rewards/margins_min": -0.01147426012903452,
|
| 495 |
+
"rewards/margins_std": 0.019380424171686172,
|
| 496 |
+
"rewards/rejected": -0.006664662156254053,
|
| 497 |
+
"step": 240
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.76,
|
| 501 |
+
"grad_norm": 2.008481756505904,
|
| 502 |
+
"learning_rate": 8.284884626103164e-08,
|
| 503 |
+
"logits/chosen": -2.817871570587158,
|
| 504 |
+
"logits/rejected": -2.786424398422241,
|
| 505 |
+
"logps/chosen": -300.6135559082031,
|
| 506 |
+
"logps/rejected": -305.0606994628906,
|
| 507 |
+
"loss": 0.6882,
|
| 508 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 509 |
+
"rewards/chosen": 0.0047633713111281395,
|
| 510 |
+
"rewards/margins": 0.009853017516434193,
|
| 511 |
+
"rewards/margins_max": 0.034555986523628235,
|
| 512 |
+
"rewards/margins_min": -0.011890431866049767,
|
| 513 |
+
"rewards/margins_std": 0.020635981112718582,
|
| 514 |
+
"rewards/rejected": -0.0050896452739834785,
|
| 515 |
+
"step": 250
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.79,
|
| 519 |
+
"grad_norm": 2.120277542804363,
|
| 520 |
+
"learning_rate": 6.409547664531733e-08,
|
| 521 |
+
"logits/chosen": -2.844655752182007,
|
| 522 |
+
"logits/rejected": -2.811575412750244,
|
| 523 |
+
"logps/chosen": -333.072265625,
|
| 524 |
+
"logps/rejected": -312.94317626953125,
|
| 525 |
+
"loss": 0.6874,
|
| 526 |
+
"rewards/accuracies": 0.762499988079071,
|
| 527 |
+
"rewards/chosen": 0.009096643887460232,
|
| 528 |
+
"rewards/margins": 0.013471168465912342,
|
| 529 |
+
"rewards/margins_max": 0.0355917289853096,
|
| 530 |
+
"rewards/margins_min": -0.005764602217823267,
|
| 531 |
+
"rewards/margins_std": 0.018313560634851456,
|
| 532 |
+
"rewards/rejected": -0.00437452457845211,
|
| 533 |
+
"step": 260
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.82,
|
| 537 |
+
"grad_norm": 2.015529778175616,
|
| 538 |
+
"learning_rate": 4.743428469705335e-08,
|
| 539 |
+
"logits/chosen": -2.7949509620666504,
|
| 540 |
+
"logits/rejected": -2.7894396781921387,
|
| 541 |
+
"logps/chosen": -303.4598693847656,
|
| 542 |
+
"logps/rejected": -308.66522216796875,
|
| 543 |
+
"loss": 0.6889,
|
| 544 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 545 |
+
"rewards/chosen": 0.003480118466541171,
|
| 546 |
+
"rewards/margins": 0.010316994972527027,
|
| 547 |
+
"rewards/margins_max": 0.033331625163555145,
|
| 548 |
+
"rewards/margins_min": -0.010707234963774681,
|
| 549 |
+
"rewards/margins_std": 0.01953895017504692,
|
| 550 |
+
"rewards/rejected": -0.006836875341832638,
|
| 551 |
+
"step": 270
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.85,
|
| 555 |
+
"grad_norm": 2.1162209644793024,
|
| 556 |
+
"learning_rate": 3.305277620188826e-08,
|
| 557 |
+
"logits/chosen": -2.844252347946167,
|
| 558 |
+
"logits/rejected": -2.8254075050354004,
|
| 559 |
+
"logps/chosen": -324.8486633300781,
|
| 560 |
+
"logps/rejected": -270.613037109375,
|
| 561 |
+
"loss": 0.6865,
|
| 562 |
+
"rewards/accuracies": 0.75,
|
| 563 |
+
"rewards/chosen": 0.0071704513393342495,
|
| 564 |
+
"rewards/margins": 0.015561411157250404,
|
| 565 |
+
"rewards/margins_max": 0.041363365948200226,
|
| 566 |
+
"rewards/margins_min": -0.010495706461369991,
|
| 567 |
+
"rewards/margins_std": 0.0231755543500185,
|
| 568 |
+
"rewards/rejected": -0.008390960283577442,
|
| 569 |
+
"step": 280
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.88,
|
| 573 |
+
"grad_norm": 1.7280929479679055,
|
| 574 |
+
"learning_rate": 2.1112801287806375e-08,
|
| 575 |
+
"logits/chosen": -2.783881187438965,
|
| 576 |
+
"logits/rejected": -2.747999668121338,
|
| 577 |
+
"logps/chosen": -273.90185546875,
|
| 578 |
+
"logps/rejected": -246.3704833984375,
|
| 579 |
+
"loss": 0.6881,
|
| 580 |
+
"rewards/accuracies": 0.699999988079071,
|
| 581 |
+
"rewards/chosen": 0.0028805662877857685,
|
| 582 |
+
"rewards/margins": 0.011384439654648304,
|
| 583 |
+
"rewards/margins_max": 0.036729536950588226,
|
| 584 |
+
"rewards/margins_min": -0.009171558544039726,
|
| 585 |
+
"rewards/margins_std": 0.021134525537490845,
|
| 586 |
+
"rewards/rejected": -0.008503873832523823,
|
| 587 |
+
"step": 290
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.91,
|
| 591 |
+
"grad_norm": 1.8137366581326853,
|
| 592 |
+
"learning_rate": 1.1748732956682023e-08,
|
| 593 |
+
"logits/chosen": -2.878770351409912,
|
| 594 |
+
"logits/rejected": -2.8104898929595947,
|
| 595 |
+
"logps/chosen": -323.51312255859375,
|
| 596 |
+
"logps/rejected": -286.44964599609375,
|
| 597 |
+
"loss": 0.6873,
|
| 598 |
+
"rewards/accuracies": 0.699999988079071,
|
| 599 |
+
"rewards/chosen": 0.0020009407307952642,
|
| 600 |
+
"rewards/margins": 0.010605795308947563,
|
| 601 |
+
"rewards/margins_max": 0.03404298424720764,
|
| 602 |
+
"rewards/margins_min": -0.010880110785365105,
|
| 603 |
+
"rewards/margins_std": 0.020114842802286148,
|
| 604 |
+
"rewards/rejected": -0.008604854345321655,
|
| 605 |
+
"step": 300
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.91,
|
| 609 |
+
"eval_logits/chosen": -2.802642822265625,
|
| 610 |
+
"eval_logits/rejected": -2.7640159130096436,
|
| 611 |
+
"eval_logps/chosen": -284.2815246582031,
|
| 612 |
+
"eval_logps/rejected": -259.24383544921875,
|
| 613 |
+
"eval_loss": 0.6889453530311584,
|
| 614 |
+
"eval_rewards/accuracies": 0.6759999990463257,
|
| 615 |
+
"eval_rewards/chosen": 0.0031190679874271154,
|
| 616 |
+
"eval_rewards/margins": 0.009768038988113403,
|
| 617 |
+
"eval_rewards/margins_max": 0.044922519475221634,
|
| 618 |
+
"eval_rewards/margins_min": -0.021590130403637886,
|
| 619 |
+
"eval_rewards/margins_std": 0.021896740421652794,
|
| 620 |
+
"eval_rewards/rejected": -0.006648970767855644,
|
| 621 |
+
"eval_runtime": 427.9336,
|
| 622 |
+
"eval_samples_per_second": 4.674,
|
| 623 |
+
"eval_steps_per_second": 0.292,
|
| 624 |
+
"step": 300
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.94,
|
| 628 |
+
"grad_norm": 1.5476563917272619,
|
| 629 |
+
"learning_rate": 5.065954844616721e-09,
|
| 630 |
+
"logits/chosen": -2.8241655826568604,
|
| 631 |
+
"logits/rejected": -2.7778286933898926,
|
| 632 |
+
"logps/chosen": -276.5940856933594,
|
| 633 |
+
"logps/rejected": -281.5748596191406,
|
| 634 |
+
"loss": 0.6885,
|
| 635 |
+
"rewards/accuracies": 0.699999988079071,
|
| 636 |
+
"rewards/chosen": 0.005276383366435766,
|
| 637 |
+
"rewards/margins": 0.010391583666205406,
|
| 638 |
+
"rewards/margins_max": 0.036186523735523224,
|
| 639 |
+
"rewards/margins_min": -0.010774780064821243,
|
| 640 |
+
"rewards/margins_std": 0.02108721435070038,
|
| 641 |
+
"rewards/rejected": -0.005115201231092215,
|
| 642 |
+
"step": 310
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.97,
|
| 646 |
+
"grad_norm": 1.9217088208809332,
|
| 647 |
+
"learning_rate": 1.1396752298723499e-09,
|
| 648 |
+
"logits/chosen": -2.8640575408935547,
|
| 649 |
+
"logits/rejected": -2.8119149208068848,
|
| 650 |
+
"logps/chosen": -249.0362548828125,
|
| 651 |
+
"logps/rejected": -258.521484375,
|
| 652 |
+
"loss": 0.6879,
|
| 653 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 654 |
+
"rewards/chosen": -0.0009104462224058807,
|
| 655 |
+
"rewards/margins": 0.008900880813598633,
|
| 656 |
+
"rewards/margins_max": 0.02946281060576439,
|
| 657 |
+
"rewards/margins_min": -0.010788346640765667,
|
| 658 |
+
"rewards/margins_std": 0.017393799498677254,
|
| 659 |
+
"rewards/rejected": -0.009811325930058956,
|
| 660 |
+
"step": 320
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 1.0,
|
| 664 |
+
"step": 329,
|
| 665 |
+
"total_flos": 0.0,
|
| 666 |
+
"train_loss": 0.6900796745323483,
|
| 667 |
+
"train_runtime": 3893.3874,
|
| 668 |
+
"train_samples_per_second": 1.352,
|
| 669 |
+
"train_steps_per_second": 0.085
|
| 670 |
+
}
|
| 671 |
+
],
|
| 672 |
+
"logging_steps": 10,
|
| 673 |
+
"max_steps": 329,
|
| 674 |
+
"num_input_tokens_seen": 0,
|
| 675 |
+
"num_train_epochs": 1,
|
| 676 |
+
"save_steps": 100,
|
| 677 |
+
"total_flos": 0.0,
|
| 678 |
+
"train_batch_size": 4,
|
| 679 |
+
"trial_name": null,
|
| 680 |
+
"trial_params": null
|
| 681 |
+
}
|