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-uf-ours-5e-7-epoch1
|
| 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-uf-ours-5e-7-epoch1
|
| 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: 0.6854
|
| 22 |
+
- Rewards/chosen: 0.0408
|
| 23 |
+
- Rewards/rejected: 0.0192
|
| 24 |
+
- Rewards/accuracies: 0.5900
|
| 25 |
+
- Rewards/margins: 0.0216
|
| 26 |
+
- Rewards/margins Max: 0.1281
|
| 27 |
+
- Rewards/margins Min: -0.0713
|
| 28 |
+
- Rewards/margins Std: 0.0665
|
| 29 |
+
- Logps/rejected: -256.6587
|
| 30 |
+
- Logps/chosen: -280.5105
|
| 31 |
+
- Logits/rejected: -2.7434
|
| 32 |
+
- Logits/chosen: -2.7814
|
| 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 |
+
| 0.6803 | 0.28 | 100 | 0.6908 | 0.0154 | 0.0089 | 0.5800 | 0.0065 | 0.0406 | -0.0225 | 0.0210 | -257.6936 | -283.0562 | -2.7654 | -2.8039 |
|
| 70 |
+
| 0.6568 | 0.56 | 200 | 0.6869 | 0.0391 | 0.0218 | 0.5960 | 0.0173 | 0.1025 | -0.0572 | 0.0531 | -256.3947 | -280.6796 | -2.7494 | -2.7877 |
|
| 71 |
+
| 0.6398 | 0.85 | 300 | 0.6854 | 0.0408 | 0.0192 | 0.5900 | 0.0216 | 0.1281 | -0.0713 | 0.0665 | -256.6587 | -280.5105 | -2.7434 | -2.7814 |
|
| 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:a3b950be076111723e48c39785001ea761fb084a91ead0079fcfee3e15951944
|
| 3 |
size 671150064
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6630057381912017,
|
| 4 |
+
"train_runtime": 4226.0112,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 1.344,
|
| 7 |
+
"train_steps_per_second": 0.084
|
| 8 |
+
}
|
runs/Jul29_11-02-19_notebook-deployment-48-7d9b6c99-khd85/events.out.tfevents.1722251039.notebook-deployment-48-7d9b6c99-khd85.3446412.0
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 sha256:4ffd1faf33012ef5097f6c802195865e50d3f8990a080c1b89a3e85fccb025c0
|
| 3 |
+
size 39981
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6630057381912017,
|
| 4 |
+
"train_runtime": 4226.0112,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 1.344,
|
| 7 |
+
"train_steps_per_second": 0.084
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,735 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 355,
|
| 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": 1.601592868423971,
|
| 14 |
+
"learning_rate": 1.3888888888888887e-08,
|
| 15 |
+
"logits/chosen": -2.861618995666504,
|
| 16 |
+
"logits/rejected": -2.8205904960632324,
|
| 17 |
+
"logps/chosen": -271.06011962890625,
|
| 18 |
+
"logps/rejected": -211.1704559326172,
|
| 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": 9.671306242111815,
|
| 32 |
+
"learning_rate": 1.3888888888888888e-07,
|
| 33 |
+
"logits/chosen": -2.8328380584716797,
|
| 34 |
+
"logits/rejected": -2.7900264263153076,
|
| 35 |
+
"logps/chosen": -325.0357971191406,
|
| 36 |
+
"logps/rejected": -275.0066833496094,
|
| 37 |
+
"loss": 0.693,
|
| 38 |
+
"rewards/accuracies": 0.4861111044883728,
|
| 39 |
+
"rewards/chosen": 0.00014538393588736653,
|
| 40 |
+
"rewards/margins": 0.000555534556042403,
|
| 41 |
+
"rewards/margins_max": 0.003110126592218876,
|
| 42 |
+
"rewards/margins_min": -0.0015574158169329166,
|
| 43 |
+
"rewards/margins_std": 0.002125632017850876,
|
| 44 |
+
"rewards/rejected": -0.0004101507074665278,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.06,
|
| 49 |
+
"grad_norm": 1.843946802013221,
|
| 50 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 51 |
+
"logits/chosen": -2.72574782371521,
|
| 52 |
+
"logits/rejected": -2.7075164318084717,
|
| 53 |
+
"logps/chosen": -293.76934814453125,
|
| 54 |
+
"logps/rejected": -215.7692108154297,
|
| 55 |
+
"loss": 0.693,
|
| 56 |
+
"rewards/accuracies": 0.5625,
|
| 57 |
+
"rewards/chosen": 0.0005661511677317321,
|
| 58 |
+
"rewards/margins": 0.0004684908199124038,
|
| 59 |
+
"rewards/margins_max": 0.003089509904384613,
|
| 60 |
+
"rewards/margins_min": -0.0020298557355999947,
|
| 61 |
+
"rewards/margins_std": 0.0022844516206532717,
|
| 62 |
+
"rewards/rejected": 9.76602896116674e-05,
|
| 63 |
+
"step": 20
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.08,
|
| 67 |
+
"grad_norm": 2.2773042809548794,
|
| 68 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 69 |
+
"logits/chosen": -2.8187897205352783,
|
| 70 |
+
"logits/rejected": -2.749154567718506,
|
| 71 |
+
"logps/chosen": -303.4768371582031,
|
| 72 |
+
"logps/rejected": -232.51852416992188,
|
| 73 |
+
"loss": 0.6921,
|
| 74 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 75 |
+
"rewards/chosen": 0.0022904500365257263,
|
| 76 |
+
"rewards/margins": 0.002333464566618204,
|
| 77 |
+
"rewards/margins_max": 0.0058710528537631035,
|
| 78 |
+
"rewards/margins_min": -0.0009718179935589433,
|
| 79 |
+
"rewards/margins_std": 0.003063285956159234,
|
| 80 |
+
"rewards/rejected": -4.3014533730456606e-05,
|
| 81 |
+
"step": 30
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.11,
|
| 85 |
+
"grad_norm": 1.731204115607293,
|
| 86 |
+
"learning_rate": 4.998060489154965e-07,
|
| 87 |
+
"logits/chosen": -2.8424630165100098,
|
| 88 |
+
"logits/rejected": -2.762080669403076,
|
| 89 |
+
"logps/chosen": -277.14581298828125,
|
| 90 |
+
"logps/rejected": -225.66372680664062,
|
| 91 |
+
"loss": 0.6915,
|
| 92 |
+
"rewards/accuracies": 0.699999988079071,
|
| 93 |
+
"rewards/chosen": 0.003388203214854002,
|
| 94 |
+
"rewards/margins": 0.002678563352674246,
|
| 95 |
+
"rewards/margins_max": 0.006867862306535244,
|
| 96 |
+
"rewards/margins_min": -0.0013280096463859081,
|
| 97 |
+
"rewards/margins_std": 0.0036745709367096424,
|
| 98 |
+
"rewards/rejected": 0.0007096402114257216,
|
| 99 |
+
"step": 40
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.14,
|
| 103 |
+
"grad_norm": 2.204111297815784,
|
| 104 |
+
"learning_rate": 4.976275538042932e-07,
|
| 105 |
+
"logits/chosen": -2.81162691116333,
|
| 106 |
+
"logits/rejected": -2.739978551864624,
|
| 107 |
+
"logps/chosen": -274.8628845214844,
|
| 108 |
+
"logps/rejected": -233.8441162109375,
|
| 109 |
+
"loss": 0.6899,
|
| 110 |
+
"rewards/accuracies": 0.887499988079071,
|
| 111 |
+
"rewards/chosen": 0.007072491105645895,
|
| 112 |
+
"rewards/margins": 0.006740429904311895,
|
| 113 |
+
"rewards/margins_max": 0.015335154719650745,
|
| 114 |
+
"rewards/margins_min": 0.0001096004998544231,
|
| 115 |
+
"rewards/margins_std": 0.0068354858085513115,
|
| 116 |
+
"rewards/rejected": 0.0003320608229842037,
|
| 117 |
+
"step": 50
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.17,
|
| 121 |
+
"grad_norm": 2.4036561931966838,
|
| 122 |
+
"learning_rate": 4.930493069997119e-07,
|
| 123 |
+
"logits/chosen": -2.7489891052246094,
|
| 124 |
+
"logits/rejected": -2.709763288497925,
|
| 125 |
+
"logps/chosen": -312.48358154296875,
|
| 126 |
+
"logps/rejected": -262.7074890136719,
|
| 127 |
+
"loss": 0.688,
|
| 128 |
+
"rewards/accuracies": 0.875,
|
| 129 |
+
"rewards/chosen": 0.011560519225895405,
|
| 130 |
+
"rewards/margins": 0.010920424945652485,
|
| 131 |
+
"rewards/margins_max": 0.021653519943356514,
|
| 132 |
+
"rewards/margins_min": 0.0011082137934863567,
|
| 133 |
+
"rewards/margins_std": 0.009159913286566734,
|
| 134 |
+
"rewards/rejected": 0.000640094222035259,
|
| 135 |
+
"step": 60
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.2,
|
| 139 |
+
"grad_norm": 1.769456739730609,
|
| 140 |
+
"learning_rate": 4.861156761634013e-07,
|
| 141 |
+
"logits/chosen": -2.783851146697998,
|
| 142 |
+
"logits/rejected": -2.7219314575195312,
|
| 143 |
+
"logps/chosen": -321.22711181640625,
|
| 144 |
+
"logps/rejected": -235.4314422607422,
|
| 145 |
+
"loss": 0.6859,
|
| 146 |
+
"rewards/accuracies": 0.875,
|
| 147 |
+
"rewards/chosen": 0.016341406852006912,
|
| 148 |
+
"rewards/margins": 0.014239917509257793,
|
| 149 |
+
"rewards/margins_max": 0.031113797798752785,
|
| 150 |
+
"rewards/margins_min": 0.0018715148326009512,
|
| 151 |
+
"rewards/margins_std": 0.013196480460464954,
|
| 152 |
+
"rewards/rejected": 0.0021014874801039696,
|
| 153 |
+
"step": 70
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.23,
|
| 157 |
+
"grad_norm": 1.861795596875086,
|
| 158 |
+
"learning_rate": 4.768938549177392e-07,
|
| 159 |
+
"logits/chosen": -2.832259178161621,
|
| 160 |
+
"logits/rejected": -2.775381565093994,
|
| 161 |
+
"logps/chosen": -319.2157897949219,
|
| 162 |
+
"logps/rejected": -286.1940002441406,
|
| 163 |
+
"loss": 0.6844,
|
| 164 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 165 |
+
"rewards/chosen": 0.020546909421682358,
|
| 166 |
+
"rewards/margins": 0.01869961991906166,
|
| 167 |
+
"rewards/margins_max": 0.04271908476948738,
|
| 168 |
+
"rewards/margins_min": 0.0003298601950518787,
|
| 169 |
+
"rewards/margins_std": 0.01946866512298584,
|
| 170 |
+
"rewards/rejected": 0.0018472913652658463,
|
| 171 |
+
"step": 80
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.25,
|
| 175 |
+
"grad_norm": 2.556237733247233,
|
| 176 |
+
"learning_rate": 4.654732116743193e-07,
|
| 177 |
+
"logits/chosen": -2.7664496898651123,
|
| 178 |
+
"logits/rejected": -2.7240653038024902,
|
| 179 |
+
"logps/chosen": -273.2864990234375,
|
| 180 |
+
"logps/rejected": -198.71231079101562,
|
| 181 |
+
"loss": 0.6806,
|
| 182 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 183 |
+
"rewards/chosen": 0.02708309330046177,
|
| 184 |
+
"rewards/margins": 0.02433420717716217,
|
| 185 |
+
"rewards/margins_max": 0.052233658730983734,
|
| 186 |
+
"rewards/margins_min": 0.0035423454828560352,
|
| 187 |
+
"rewards/margins_std": 0.022480051964521408,
|
| 188 |
+
"rewards/rejected": 0.002748883794993162,
|
| 189 |
+
"step": 90
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.28,
|
| 193 |
+
"grad_norm": 1.897335973090835,
|
| 194 |
+
"learning_rate": 4.519644235671752e-07,
|
| 195 |
+
"logits/chosen": -2.8063488006591797,
|
| 196 |
+
"logits/rejected": -2.768160581588745,
|
| 197 |
+
"logps/chosen": -278.94610595703125,
|
| 198 |
+
"logps/rejected": -261.92315673828125,
|
| 199 |
+
"loss": 0.6803,
|
| 200 |
+
"rewards/accuracies": 0.862500011920929,
|
| 201 |
+
"rewards/chosen": 0.024656174704432487,
|
| 202 |
+
"rewards/margins": 0.022013681009411812,
|
| 203 |
+
"rewards/margins_max": 0.04718270152807236,
|
| 204 |
+
"rewards/margins_min": 0.0019731284119188786,
|
| 205 |
+
"rewards/margins_std": 0.020801713690161705,
|
| 206 |
+
"rewards/rejected": 0.0026424946263432503,
|
| 207 |
+
"step": 100
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.28,
|
| 211 |
+
"eval_logits/chosen": -2.803894519805908,
|
| 212 |
+
"eval_logits/rejected": -2.7653636932373047,
|
| 213 |
+
"eval_logps/chosen": -283.0562438964844,
|
| 214 |
+
"eval_logps/rejected": -257.6935729980469,
|
| 215 |
+
"eval_loss": 0.6907610297203064,
|
| 216 |
+
"eval_rewards/accuracies": 0.5799999833106995,
|
| 217 |
+
"eval_rewards/chosen": 0.01537203136831522,
|
| 218 |
+
"eval_rewards/margins": 0.00651828246191144,
|
| 219 |
+
"eval_rewards/margins_max": 0.04056054726243019,
|
| 220 |
+
"eval_rewards/margins_min": -0.02254258282482624,
|
| 221 |
+
"eval_rewards/margins_std": 0.020977748557925224,
|
| 222 |
+
"eval_rewards/rejected": 0.008853749372065067,
|
| 223 |
+
"eval_runtime": 431.2136,
|
| 224 |
+
"eval_samples_per_second": 4.638,
|
| 225 |
+
"eval_steps_per_second": 0.29,
|
| 226 |
+
"step": 100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.31,
|
| 230 |
+
"grad_norm": 2.350716473544898,
|
| 231 |
+
"learning_rate": 4.364984038837727e-07,
|
| 232 |
+
"logits/chosen": -2.8714425563812256,
|
| 233 |
+
"logits/rejected": -2.776231527328491,
|
| 234 |
+
"logps/chosen": -370.3124084472656,
|
| 235 |
+
"logps/rejected": -286.1990051269531,
|
| 236 |
+
"loss": 0.6739,
|
| 237 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 238 |
+
"rewards/chosen": 0.042933039367198944,
|
| 239 |
+
"rewards/margins": 0.04058496281504631,
|
| 240 |
+
"rewards/margins_max": 0.07831507176160812,
|
| 241 |
+
"rewards/margins_min": 0.009233040735125542,
|
| 242 |
+
"rewards/margins_std": 0.03092753328382969,
|
| 243 |
+
"rewards/rejected": 0.0023480760864913464,
|
| 244 |
+
"step": 110
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.34,
|
| 248 |
+
"grad_norm": 2.0267120141468924,
|
| 249 |
+
"learning_rate": 4.1922503338800447e-07,
|
| 250 |
+
"logits/chosen": -2.840080976486206,
|
| 251 |
+
"logits/rejected": -2.7854738235473633,
|
| 252 |
+
"logps/chosen": -343.77545166015625,
|
| 253 |
+
"logps/rejected": -265.7751159667969,
|
| 254 |
+
"loss": 0.6727,
|
| 255 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 256 |
+
"rewards/chosen": 0.046107616275548935,
|
| 257 |
+
"rewards/margins": 0.04191668704152107,
|
| 258 |
+
"rewards/margins_max": 0.0861145481467247,
|
| 259 |
+
"rewards/margins_min": 0.010150762274861336,
|
| 260 |
+
"rewards/margins_std": 0.035071179270744324,
|
| 261 |
+
"rewards/rejected": 0.004190927837044001,
|
| 262 |
+
"step": 120
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.37,
|
| 266 |
+
"grad_norm": 1.7790024615351698,
|
| 267 |
+
"learning_rate": 4.003117078299021e-07,
|
| 268 |
+
"logits/chosen": -2.8337936401367188,
|
| 269 |
+
"logits/rejected": -2.751962184906006,
|
| 270 |
+
"logps/chosen": -377.0141906738281,
|
| 271 |
+
"logps/rejected": -300.07891845703125,
|
| 272 |
+
"loss": 0.6685,
|
| 273 |
+
"rewards/accuracies": 0.949999988079071,
|
| 274 |
+
"rewards/chosen": 0.05350027605891228,
|
| 275 |
+
"rewards/margins": 0.04998067393898964,
|
| 276 |
+
"rewards/margins_max": 0.09481509029865265,
|
| 277 |
+
"rewards/margins_min": 0.008272581733763218,
|
| 278 |
+
"rewards/margins_std": 0.03825851529836655,
|
| 279 |
+
"rewards/rejected": 0.0035195990931242704,
|
| 280 |
+
"step": 130
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.39,
|
| 284 |
+
"grad_norm": 1.4853892826304387,
|
| 285 |
+
"learning_rate": 3.799417157181075e-07,
|
| 286 |
+
"logits/chosen": -2.786635398864746,
|
| 287 |
+
"logits/rejected": -2.7476704120635986,
|
| 288 |
+
"logps/chosen": -312.93548583984375,
|
| 289 |
+
"logps/rejected": -269.5070495605469,
|
| 290 |
+
"loss": 0.6703,
|
| 291 |
+
"rewards/accuracies": 0.925000011920929,
|
| 292 |
+
"rewards/chosen": 0.05478551238775253,
|
| 293 |
+
"rewards/margins": 0.05136098712682724,
|
| 294 |
+
"rewards/margins_max": 0.10997221618890762,
|
| 295 |
+
"rewards/margins_min": 0.007977863773703575,
|
| 296 |
+
"rewards/margins_std": 0.04762490838766098,
|
| 297 |
+
"rewards/rejected": 0.003424527822062373,
|
| 298 |
+
"step": 140
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.42,
|
| 302 |
+
"grad_norm": 1.9607644783681,
|
| 303 |
+
"learning_rate": 3.583124620760659e-07,
|
| 304 |
+
"logits/chosen": -2.788800001144409,
|
| 305 |
+
"logits/rejected": -2.748765468597412,
|
| 306 |
+
"logps/chosen": -283.2447814941406,
|
| 307 |
+
"logps/rejected": -212.7663116455078,
|
| 308 |
+
"loss": 0.6666,
|
| 309 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 310 |
+
"rewards/chosen": 0.05440182611346245,
|
| 311 |
+
"rewards/margins": 0.051237184554338455,
|
| 312 |
+
"rewards/margins_max": 0.10327558219432831,
|
| 313 |
+
"rewards/margins_min": 0.00882364995777607,
|
| 314 |
+
"rewards/margins_std": 0.04341956600546837,
|
| 315 |
+
"rewards/rejected": 0.0031646445859223604,
|
| 316 |
+
"step": 150
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.45,
|
| 320 |
+
"grad_norm": 1.6524576639358914,
|
| 321 |
+
"learning_rate": 3.356335553954679e-07,
|
| 322 |
+
"logits/chosen": -2.766443967819214,
|
| 323 |
+
"logits/rejected": -2.712019681930542,
|
| 324 |
+
"logps/chosen": -294.170654296875,
|
| 325 |
+
"logps/rejected": -233.78359985351562,
|
| 326 |
+
"loss": 0.664,
|
| 327 |
+
"rewards/accuracies": 0.9375,
|
| 328 |
+
"rewards/chosen": 0.06376218795776367,
|
| 329 |
+
"rewards/margins": 0.05766482278704643,
|
| 330 |
+
"rewards/margins_max": 0.11248371750116348,
|
| 331 |
+
"rewards/margins_min": 0.00960196740925312,
|
| 332 |
+
"rewards/margins_std": 0.046693481504917145,
|
| 333 |
+
"rewards/rejected": 0.0060973623767495155,
|
| 334 |
+
"step": 160
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.48,
|
| 338 |
+
"grad_norm": 1.829740022838758,
|
| 339 |
+
"learning_rate": 3.121247763262235e-07,
|
| 340 |
+
"logits/chosen": -2.8182921409606934,
|
| 341 |
+
"logits/rejected": -2.7600901126861572,
|
| 342 |
+
"logps/chosen": -322.2862854003906,
|
| 343 |
+
"logps/rejected": -296.15496826171875,
|
| 344 |
+
"loss": 0.6602,
|
| 345 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 346 |
+
"rewards/chosen": 0.07335338741540909,
|
| 347 |
+
"rewards/margins": 0.07224629819393158,
|
| 348 |
+
"rewards/margins_max": 0.13527169823646545,
|
| 349 |
+
"rewards/margins_min": 0.014563126489520073,
|
| 350 |
+
"rewards/margins_std": 0.054327212274074554,
|
| 351 |
+
"rewards/rejected": 0.001107088173739612,
|
| 352 |
+
"step": 170
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.51,
|
| 356 |
+
"grad_norm": 1.954322630861037,
|
| 357 |
+
"learning_rate": 2.880139477883347e-07,
|
| 358 |
+
"logits/chosen": -2.8129706382751465,
|
| 359 |
+
"logits/rejected": -2.72513484954834,
|
| 360 |
+
"logps/chosen": -324.20794677734375,
|
| 361 |
+
"logps/rejected": -293.87493896484375,
|
| 362 |
+
"loss": 0.6567,
|
| 363 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 364 |
+
"rewards/chosen": 0.07692877948284149,
|
| 365 |
+
"rewards/margins": 0.07382867485284805,
|
| 366 |
+
"rewards/margins_max": 0.14840444922447205,
|
| 367 |
+
"rewards/margins_min": 0.021513966843485832,
|
| 368 |
+
"rewards/margins_std": 0.05711384862661362,
|
| 369 |
+
"rewards/rejected": 0.0031001190654933453,
|
| 370 |
+
"step": 180
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.54,
|
| 374 |
+
"grad_norm": 2.352020928698477,
|
| 375 |
+
"learning_rate": 2.635347271463544e-07,
|
| 376 |
+
"logits/chosen": -2.7723231315612793,
|
| 377 |
+
"logits/rejected": -2.6797118186950684,
|
| 378 |
+
"logps/chosen": -313.4515686035156,
|
| 379 |
+
"logps/rejected": -237.2364501953125,
|
| 380 |
+
"loss": 0.6551,
|
| 381 |
+
"rewards/accuracies": 0.925000011920929,
|
| 382 |
+
"rewards/chosen": 0.08509302884340286,
|
| 383 |
+
"rewards/margins": 0.08258074522018433,
|
| 384 |
+
"rewards/margins_max": 0.16938893496990204,
|
| 385 |
+
"rewards/margins_min": 0.015894772484898567,
|
| 386 |
+
"rewards/margins_std": 0.07122750580310822,
|
| 387 |
+
"rewards/rejected": 0.002512281993404031,
|
| 388 |
+
"step": 190
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.56,
|
| 392 |
+
"grad_norm": 2.051556424366539,
|
| 393 |
+
"learning_rate": 2.3892434184240534e-07,
|
| 394 |
+
"logits/chosen": -2.8492965698242188,
|
| 395 |
+
"logits/rejected": -2.7720444202423096,
|
| 396 |
+
"logps/chosen": -330.9425048828125,
|
| 397 |
+
"logps/rejected": -265.1010437011719,
|
| 398 |
+
"loss": 0.6568,
|
| 399 |
+
"rewards/accuracies": 0.9375,
|
| 400 |
+
"rewards/chosen": 0.08590307831764221,
|
| 401 |
+
"rewards/margins": 0.0809393972158432,
|
| 402 |
+
"rewards/margins_max": 0.16236025094985962,
|
| 403 |
+
"rewards/margins_min": 0.016894496977329254,
|
| 404 |
+
"rewards/margins_std": 0.06359551846981049,
|
| 405 |
+
"rewards/rejected": 0.004963680170476437,
|
| 406 |
+
"step": 200
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.56,
|
| 410 |
+
"eval_logits/chosen": -2.787705421447754,
|
| 411 |
+
"eval_logits/rejected": -2.749375343322754,
|
| 412 |
+
"eval_logps/chosen": -280.6795959472656,
|
| 413 |
+
"eval_logps/rejected": -256.39471435546875,
|
| 414 |
+
"eval_loss": 0.6869410276412964,
|
| 415 |
+
"eval_rewards/accuracies": 0.5960000157356262,
|
| 416 |
+
"eval_rewards/chosen": 0.03913822025060654,
|
| 417 |
+
"eval_rewards/margins": 0.01729637198150158,
|
| 418 |
+
"eval_rewards/margins_max": 0.1024816706776619,
|
| 419 |
+
"eval_rewards/margins_min": -0.05720587447285652,
|
| 420 |
+
"eval_rewards/margins_std": 0.05308415740728378,
|
| 421 |
+
"eval_rewards/rejected": 0.021841851994395256,
|
| 422 |
+
"eval_runtime": 429.0815,
|
| 423 |
+
"eval_samples_per_second": 4.661,
|
| 424 |
+
"eval_steps_per_second": 0.291,
|
| 425 |
+
"step": 200
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.59,
|
| 429 |
+
"grad_norm": 1.9346913930945893,
|
| 430 |
+
"learning_rate": 2.1442129043167873e-07,
|
| 431 |
+
"logits/chosen": -2.7502646446228027,
|
| 432 |
+
"logits/rejected": -2.7227656841278076,
|
| 433 |
+
"logps/chosen": -310.6435852050781,
|
| 434 |
+
"logps/rejected": -256.34246826171875,
|
| 435 |
+
"loss": 0.6507,
|
| 436 |
+
"rewards/accuracies": 0.949999988079071,
|
| 437 |
+
"rewards/chosen": 0.09540757536888123,
|
| 438 |
+
"rewards/margins": 0.09458459913730621,
|
| 439 |
+
"rewards/margins_max": 0.1914738416671753,
|
| 440 |
+
"rewards/margins_min": 0.01557054091244936,
|
| 441 |
+
"rewards/margins_std": 0.0801827609539032,
|
| 442 |
+
"rewards/rejected": 0.0008229793747887015,
|
| 443 |
+
"step": 210
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.62,
|
| 447 |
+
"grad_norm": 1.8397256282784018,
|
| 448 |
+
"learning_rate": 1.9026303129961048e-07,
|
| 449 |
+
"logits/chosen": -2.849534511566162,
|
| 450 |
+
"logits/rejected": -2.7543559074401855,
|
| 451 |
+
"logps/chosen": -343.22222900390625,
|
| 452 |
+
"logps/rejected": -274.1776428222656,
|
| 453 |
+
"loss": 0.6479,
|
| 454 |
+
"rewards/accuracies": 0.9375,
|
| 455 |
+
"rewards/chosen": 0.10234014689922333,
|
| 456 |
+
"rewards/margins": 0.10084100067615509,
|
| 457 |
+
"rewards/margins_max": 0.1773754209280014,
|
| 458 |
+
"rewards/margins_min": 0.02473507821559906,
|
| 459 |
+
"rewards/margins_std": 0.06906573474407196,
|
| 460 |
+
"rewards/rejected": 0.001499147154390812,
|
| 461 |
+
"step": 220
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.65,
|
| 465 |
+
"grad_norm": 1.9517251453168138,
|
| 466 |
+
"learning_rate": 1.6668368145931396e-07,
|
| 467 |
+
"logits/chosen": -2.8495993614196777,
|
| 468 |
+
"logits/rejected": -2.756133556365967,
|
| 469 |
+
"logps/chosen": -361.55810546875,
|
| 470 |
+
"logps/rejected": -262.0633850097656,
|
| 471 |
+
"loss": 0.6422,
|
| 472 |
+
"rewards/accuracies": 0.949999988079071,
|
| 473 |
+
"rewards/chosen": 0.1011759415268898,
|
| 474 |
+
"rewards/margins": 0.10087559372186661,
|
| 475 |
+
"rewards/margins_max": 0.19324660301208496,
|
| 476 |
+
"rewards/margins_min": 0.022996682673692703,
|
| 477 |
+
"rewards/margins_std": 0.07676897943019867,
|
| 478 |
+
"rewards/rejected": 0.0003003578167408705,
|
| 479 |
+
"step": 230
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.68,
|
| 483 |
+
"grad_norm": 1.5814942001733312,
|
| 484 |
+
"learning_rate": 1.4391174773015834e-07,
|
| 485 |
+
"logits/chosen": -2.7951114177703857,
|
| 486 |
+
"logits/rejected": -2.724926233291626,
|
| 487 |
+
"logps/chosen": -316.5010070800781,
|
| 488 |
+
"logps/rejected": -283.17291259765625,
|
| 489 |
+
"loss": 0.6514,
|
| 490 |
+
"rewards/accuracies": 0.925000011920929,
|
| 491 |
+
"rewards/chosen": 0.0829603299498558,
|
| 492 |
+
"rewards/margins": 0.08294094353914261,
|
| 493 |
+
"rewards/margins_max": 0.16664841771125793,
|
| 494 |
+
"rewards/margins_min": 0.015887869521975517,
|
| 495 |
+
"rewards/margins_std": 0.06911344826221466,
|
| 496 |
+
"rewards/rejected": 1.939358116942458e-05,
|
| 497 |
+
"step": 240
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.7,
|
| 501 |
+
"grad_norm": 1.9477304039773469,
|
| 502 |
+
"learning_rate": 1.2216791228457775e-07,
|
| 503 |
+
"logits/chosen": -2.795609951019287,
|
| 504 |
+
"logits/rejected": -2.7185869216918945,
|
| 505 |
+
"logps/chosen": -302.67822265625,
|
| 506 |
+
"logps/rejected": -251.4832000732422,
|
| 507 |
+
"loss": 0.6494,
|
| 508 |
+
"rewards/accuracies": 0.949999988079071,
|
| 509 |
+
"rewards/chosen": 0.0860290601849556,
|
| 510 |
+
"rewards/margins": 0.090418241918087,
|
| 511 |
+
"rewards/margins_max": 0.18044039607048035,
|
| 512 |
+
"rewards/margins_min": 0.016120051965117455,
|
| 513 |
+
"rewards/margins_std": 0.07592571526765823,
|
| 514 |
+
"rewards/rejected": -0.0043891845270991325,
|
| 515 |
+
"step": 250
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.73,
|
| 519 |
+
"grad_norm": 1.7138528589589788,
|
| 520 |
+
"learning_rate": 1.0166289402331391e-07,
|
| 521 |
+
"logits/chosen": -2.8433239459991455,
|
| 522 |
+
"logits/rejected": -2.760486125946045,
|
| 523 |
+
"logps/chosen": -282.75128173828125,
|
| 524 |
+
"logps/rejected": -259.15283203125,
|
| 525 |
+
"loss": 0.6432,
|
| 526 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 527 |
+
"rewards/chosen": 0.09342271834611893,
|
| 528 |
+
"rewards/margins": 0.09564103186130524,
|
| 529 |
+
"rewards/margins_max": 0.19372263550758362,
|
| 530 |
+
"rewards/margins_min": 0.020370684564113617,
|
| 531 |
+
"rewards/margins_std": 0.08056275546550751,
|
| 532 |
+
"rewards/rejected": -0.0022183198016136885,
|
| 533 |
+
"step": 260
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.76,
|
| 537 |
+
"grad_norm": 2.0016114180555578,
|
| 538 |
+
"learning_rate": 8.259540650444734e-08,
|
| 539 |
+
"logits/chosen": -2.7903430461883545,
|
| 540 |
+
"logits/rejected": -2.7373478412628174,
|
| 541 |
+
"logps/chosen": -300.1326599121094,
|
| 542 |
+
"logps/rejected": -262.3350830078125,
|
| 543 |
+
"loss": 0.6496,
|
| 544 |
+
"rewards/accuracies": 0.925000011920929,
|
| 545 |
+
"rewards/chosen": 0.08982918411493301,
|
| 546 |
+
"rewards/margins": 0.08818400651216507,
|
| 547 |
+
"rewards/margins_max": 0.16805505752563477,
|
| 548 |
+
"rewards/margins_min": 0.026935186237096786,
|
| 549 |
+
"rewards/margins_std": 0.06377857178449631,
|
| 550 |
+
"rewards/rejected": 0.0016451816773042083,
|
| 551 |
+
"step": 270
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.79,
|
| 555 |
+
"grad_norm": 1.7516715377367251,
|
| 556 |
+
"learning_rate": 6.515023221586721e-08,
|
| 557 |
+
"logits/chosen": -2.766188144683838,
|
| 558 |
+
"logits/rejected": -2.7269351482391357,
|
| 559 |
+
"logps/chosen": -294.2494812011719,
|
| 560 |
+
"logps/rejected": -272.6824951171875,
|
| 561 |
+
"loss": 0.648,
|
| 562 |
+
"rewards/accuracies": 0.9624999761581421,
|
| 563 |
+
"rewards/chosen": 0.09207499772310257,
|
| 564 |
+
"rewards/margins": 0.09292837232351303,
|
| 565 |
+
"rewards/margins_max": 0.17264562845230103,
|
| 566 |
+
"rewards/margins_min": 0.023857418447732925,
|
| 567 |
+
"rewards/margins_std": 0.06797114759683609,
|
| 568 |
+
"rewards/rejected": -0.0008533738437108696,
|
| 569 |
+
"step": 280
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.82,
|
| 573 |
+
"grad_norm": 1.7538431067028422,
|
| 574 |
+
"learning_rate": 4.949643185335287e-08,
|
| 575 |
+
"logits/chosen": -2.771496295928955,
|
| 576 |
+
"logits/rejected": -2.721856117248535,
|
| 577 |
+
"logps/chosen": -278.96405029296875,
|
| 578 |
+
"logps/rejected": -264.3961181640625,
|
| 579 |
+
"loss": 0.653,
|
| 580 |
+
"rewards/accuracies": 0.887499988079071,
|
| 581 |
+
"rewards/chosen": 0.07952173054218292,
|
| 582 |
+
"rewards/margins": 0.08330051600933075,
|
| 583 |
+
"rewards/margins_max": 0.18094156682491302,
|
| 584 |
+
"rewards/margins_min": 0.007182260509580374,
|
| 585 |
+
"rewards/margins_std": 0.07998403161764145,
|
| 586 |
+
"rewards/rejected": -0.0037787805777043104,
|
| 587 |
+
"step": 290
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.85,
|
| 591 |
+
"grad_norm": 1.9243414334438866,
|
| 592 |
+
"learning_rate": 3.578570595810274e-08,
|
| 593 |
+
"logits/chosen": -2.8379387855529785,
|
| 594 |
+
"logits/rejected": -2.7606096267700195,
|
| 595 |
+
"logps/chosen": -335.54931640625,
|
| 596 |
+
"logps/rejected": -289.7465515136719,
|
| 597 |
+
"loss": 0.6398,
|
| 598 |
+
"rewards/accuracies": 0.9375,
|
| 599 |
+
"rewards/chosen": 0.10963740199804306,
|
| 600 |
+
"rewards/margins": 0.11276860535144806,
|
| 601 |
+
"rewards/margins_max": 0.21979229152202606,
|
| 602 |
+
"rewards/margins_min": 0.021601524204015732,
|
| 603 |
+
"rewards/margins_std": 0.08902540057897568,
|
| 604 |
+
"rewards/rejected": -0.0031312075443565845,
|
| 605 |
+
"step": 300
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.85,
|
| 609 |
+
"eval_logits/chosen": -2.781425952911377,
|
| 610 |
+
"eval_logits/rejected": -2.743434190750122,
|
| 611 |
+
"eval_logps/chosen": -280.510498046875,
|
| 612 |
+
"eval_logps/rejected": -256.65869140625,
|
| 613 |
+
"eval_loss": 0.6854214668273926,
|
| 614 |
+
"eval_rewards/accuracies": 0.5899999737739563,
|
| 615 |
+
"eval_rewards/chosen": 0.04082921892404556,
|
| 616 |
+
"eval_rewards/margins": 0.021626880392432213,
|
| 617 |
+
"eval_rewards/margins_max": 0.12814800441265106,
|
| 618 |
+
"eval_rewards/margins_min": -0.07126501202583313,
|
| 619 |
+
"eval_rewards/margins_std": 0.06649063527584076,
|
| 620 |
+
"eval_rewards/rejected": 0.0192023366689682,
|
| 621 |
+
"eval_runtime": 428.7677,
|
| 622 |
+
"eval_samples_per_second": 4.665,
|
| 623 |
+
"eval_steps_per_second": 0.292,
|
| 624 |
+
"step": 300
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.87,
|
| 628 |
+
"grad_norm": 1.7722376322831996,
|
| 629 |
+
"learning_rate": 2.415092479103503e-08,
|
| 630 |
+
"logits/chosen": -2.835430860519409,
|
| 631 |
+
"logits/rejected": -2.740483522415161,
|
| 632 |
+
"logps/chosen": -287.6938171386719,
|
| 633 |
+
"logps/rejected": -214.74563598632812,
|
| 634 |
+
"loss": 0.64,
|
| 635 |
+
"rewards/accuracies": 0.987500011920929,
|
| 636 |
+
"rewards/chosen": 0.10122032463550568,
|
| 637 |
+
"rewards/margins": 0.10643725097179413,
|
| 638 |
+
"rewards/margins_max": 0.21550527215003967,
|
| 639 |
+
"rewards/margins_min": 0.02140904776751995,
|
| 640 |
+
"rewards/margins_std": 0.08673234283924103,
|
| 641 |
+
"rewards/rejected": -0.005216927733272314,
|
| 642 |
+
"step": 310
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.9,
|
| 646 |
+
"grad_norm": 1.8570363058712267,
|
| 647 |
+
"learning_rate": 1.4704840690808656e-08,
|
| 648 |
+
"logits/chosen": -2.80118727684021,
|
| 649 |
+
"logits/rejected": -2.743530035018921,
|
| 650 |
+
"logps/chosen": -298.8092956542969,
|
| 651 |
+
"logps/rejected": -260.87152099609375,
|
| 652 |
+
"loss": 0.6445,
|
| 653 |
+
"rewards/accuracies": 0.9375,
|
| 654 |
+
"rewards/chosen": 0.09463869780302048,
|
| 655 |
+
"rewards/margins": 0.09665326029062271,
|
| 656 |
+
"rewards/margins_max": 0.20018891990184784,
|
| 657 |
+
"rewards/margins_min": 0.01689526066184044,
|
| 658 |
+
"rewards/margins_std": 0.08293472975492477,
|
| 659 |
+
"rewards/rejected": -0.0020145513117313385,
|
| 660 |
+
"step": 320
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.93,
|
| 664 |
+
"grad_norm": 1.9373896782753002,
|
| 665 |
+
"learning_rate": 7.538995394063995e-09,
|
| 666 |
+
"logits/chosen": -2.8751907348632812,
|
| 667 |
+
"logits/rejected": -2.7946133613586426,
|
| 668 |
+
"logps/chosen": -343.7359619140625,
|
| 669 |
+
"logps/rejected": -267.89105224609375,
|
| 670 |
+
"loss": 0.6408,
|
| 671 |
+
"rewards/accuracies": 0.987500011920929,
|
| 672 |
+
"rewards/chosen": 0.10583887249231339,
|
| 673 |
+
"rewards/margins": 0.11535302549600601,
|
| 674 |
+
"rewards/margins_max": 0.20861783623695374,
|
| 675 |
+
"rewards/margins_min": 0.029701709747314453,
|
| 676 |
+
"rewards/margins_std": 0.08170617371797562,
|
| 677 |
+
"rewards/rejected": -0.009514153003692627,
|
| 678 |
+
"step": 330
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.96,
|
| 682 |
+
"grad_norm": 1.9173076290433462,
|
| 683 |
+
"learning_rate": 2.7228329070159705e-09,
|
| 684 |
+
"logits/chosen": -2.762208938598633,
|
| 685 |
+
"logits/rejected": -2.710221767425537,
|
| 686 |
+
"logps/chosen": -289.7626037597656,
|
| 687 |
+
"logps/rejected": -251.68887329101562,
|
| 688 |
+
"loss": 0.6385,
|
| 689 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 690 |
+
"rewards/chosen": 0.10082916170358658,
|
| 691 |
+
"rewards/margins": 0.10602158308029175,
|
| 692 |
+
"rewards/margins_max": 0.22432371973991394,
|
| 693 |
+
"rewards/margins_min": 0.02705308236181736,
|
| 694 |
+
"rewards/margins_std": 0.09002183377742767,
|
| 695 |
+
"rewards/rejected": -0.005192426033318043,
|
| 696 |
+
"step": 340
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.99,
|
| 700 |
+
"grad_norm": 5.134028042605605,
|
| 701 |
+
"learning_rate": 3.0302652553296226e-10,
|
| 702 |
+
"logits/chosen": -2.7456893920898438,
|
| 703 |
+
"logits/rejected": -2.7021021842956543,
|
| 704 |
+
"logps/chosen": -306.5563049316406,
|
| 705 |
+
"logps/rejected": -287.2024230957031,
|
| 706 |
+
"loss": 0.647,
|
| 707 |
+
"rewards/accuracies": 0.925000011920929,
|
| 708 |
+
"rewards/chosen": 0.08402810990810394,
|
| 709 |
+
"rewards/margins": 0.09153296053409576,
|
| 710 |
+
"rewards/margins_max": 0.18022149801254272,
|
| 711 |
+
"rewards/margins_min": 0.017785457894206047,
|
| 712 |
+
"rewards/margins_std": 0.07233723253011703,
|
| 713 |
+
"rewards/rejected": -0.007504852022975683,
|
| 714 |
+
"step": 350
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.0,
|
| 718 |
+
"step": 355,
|
| 719 |
+
"total_flos": 0.0,
|
| 720 |
+
"train_loss": 0.6630057381912017,
|
| 721 |
+
"train_runtime": 4226.0112,
|
| 722 |
+
"train_samples_per_second": 1.344,
|
| 723 |
+
"train_steps_per_second": 0.084
|
| 724 |
+
}
|
| 725 |
+
],
|
| 726 |
+
"logging_steps": 10,
|
| 727 |
+
"max_steps": 355,
|
| 728 |
+
"num_input_tokens_seen": 0,
|
| 729 |
+
"num_train_epochs": 1,
|
| 730 |
+
"save_steps": 100,
|
| 731 |
+
"total_flos": 0.0,
|
| 732 |
+
"train_batch_size": 4,
|
| 733 |
+
"trial_name": null,
|
| 734 |
+
"trial_params": null
|
| 735 |
+
}
|