Model save
Browse files- README.md +77 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +600 -0
README.md
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: mistralai/Mistral-7B-v0.1
|
| 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-7b-dpo-qlora
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-7b-dpo-qlora
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 1964.4768
|
| 22 |
+
- Rewards/chosen: -0.1407
|
| 23 |
+
- Rewards/rejected: -0.2359
|
| 24 |
+
- Rewards/accuracies: 0.7243
|
| 25 |
+
- Rewards/margins: 0.0952
|
| 26 |
+
- Logps/rejected: -25.2353
|
| 27 |
+
- Logps/chosen: -15.6615
|
| 28 |
+
- Logits/rejected: -6.0582
|
| 29 |
+
- Logits/chosen: -5.0283
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-06
|
| 49 |
+
- train_batch_size: 4
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 8
|
| 54 |
+
- gradient_accumulation_steps: 4
|
| 55 |
+
- total_train_batch_size: 128
|
| 56 |
+
- total_eval_batch_size: 64
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 1
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 2178.8832 | 0.3017 | 100 | 2139.6340 | -0.1014 | -0.1518 | 0.6781 | 0.0504 | -16.8316 | -11.7293 | -3.3193 | -3.1137 |
|
| 67 |
+
| 1986.1408 | 0.6033 | 200 | 1989.8781 | -0.1408 | -0.2280 | 0.7055 | 0.0872 | -24.4503 | -15.6687 | -6.1921 | -5.2910 |
|
| 68 |
+
| 2000.1098 | 0.9050 | 300 | 1964.4768 | -0.1407 | -0.2359 | 0.7243 | 0.0952 | -25.2353 | -15.6615 | -6.0582 | -5.0283 |
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
### Framework versions
|
| 72 |
+
|
| 73 |
+
- PEFT 0.12.0
|
| 74 |
+
- Transformers 4.43.3
|
| 75 |
+
- Pytorch 2.4.0+cu121
|
| 76 |
+
- Datasets 2.20.0
|
| 77 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9984917043740573,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 2150.5997450906343,
|
| 5 |
+
"train_runtime": 2663.9905,
|
| 6 |
+
"train_samples": 42421,
|
| 7 |
+
"train_samples_per_second": 15.924,
|
| 8 |
+
"train_steps_per_second": 0.124
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9984917043740573,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 2150.5997450906343,
|
| 5 |
+
"train_runtime": 2663.9905,
|
| 6 |
+
"train_samples": 42421,
|
| 7 |
+
"train_samples_per_second": 15.924,
|
| 8 |
+
"train_steps_per_second": 0.124
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,600 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9984917043740573,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 331,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0030165912518853697,
|
| 13 |
+
"grad_norm": 153.23599243164062,
|
| 14 |
+
"learning_rate": 1.4705882352941178e-07,
|
| 15 |
+
"logits/chosen": -2.375755786895752,
|
| 16 |
+
"logits/rejected": -2.402980327606201,
|
| 17 |
+
"logps/chosen": -1.5958898067474365,
|
| 18 |
+
"logps/rejected": -1.7740895748138428,
|
| 19 |
+
"loss": 2500.0,
|
| 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": 0.030165912518853696,
|
| 28 |
+
"grad_norm": 118.78604125976562,
|
| 29 |
+
"learning_rate": 1.4705882352941177e-06,
|
| 30 |
+
"logits/chosen": -2.3309760093688965,
|
| 31 |
+
"logits/rejected": -2.383087635040283,
|
| 32 |
+
"logps/chosen": -1.5815058946609497,
|
| 33 |
+
"logps/rejected": -1.6628098487854004,
|
| 34 |
+
"loss": 2499.6899,
|
| 35 |
+
"rewards/accuracies": 0.4930555522441864,
|
| 36 |
+
"rewards/chosen": 0.00026574183721095324,
|
| 37 |
+
"rewards/margins": 1.8858834664570168e-05,
|
| 38 |
+
"rewards/rejected": 0.0002468830207362771,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.06033182503770739,
|
| 43 |
+
"grad_norm": 107.57190704345703,
|
| 44 |
+
"learning_rate": 2.9411764705882355e-06,
|
| 45 |
+
"logits/chosen": -2.3686156272888184,
|
| 46 |
+
"logits/rejected": -2.4214446544647217,
|
| 47 |
+
"logps/chosen": -1.5373353958129883,
|
| 48 |
+
"logps/rejected": -1.5770070552825928,
|
| 49 |
+
"loss": 2498.7738,
|
| 50 |
+
"rewards/accuracies": 0.512499988079071,
|
| 51 |
+
"rewards/chosen": 0.001089997822418809,
|
| 52 |
+
"rewards/margins": 0.00014883658150210977,
|
| 53 |
+
"rewards/rejected": 0.0009411612409166992,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.09049773755656108,
|
| 58 |
+
"grad_norm": 116.10955810546875,
|
| 59 |
+
"learning_rate": 4.411764705882353e-06,
|
| 60 |
+
"logits/chosen": -2.4276905059814453,
|
| 61 |
+
"logits/rejected": -2.4831461906433105,
|
| 62 |
+
"logps/chosen": -1.5260236263275146,
|
| 63 |
+
"logps/rejected": -1.5350881814956665,
|
| 64 |
+
"loss": 2497.6447,
|
| 65 |
+
"rewards/accuracies": 0.59375,
|
| 66 |
+
"rewards/chosen": 0.0010487588588148355,
|
| 67 |
+
"rewards/margins": 0.00025444405036978424,
|
| 68 |
+
"rewards/rejected": 0.0007943148957565427,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.12066365007541478,
|
| 73 |
+
"grad_norm": 207.2774200439453,
|
| 74 |
+
"learning_rate": 4.994966691179712e-06,
|
| 75 |
+
"logits/chosen": -2.456132411956787,
|
| 76 |
+
"logits/rejected": -2.4862520694732666,
|
| 77 |
+
"logps/chosen": -1.5139974355697632,
|
| 78 |
+
"logps/rejected": -1.6103894710540771,
|
| 79 |
+
"loss": 2494.1223,
|
| 80 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 81 |
+
"rewards/chosen": 0.00042313727317377925,
|
| 82 |
+
"rewards/margins": 0.0003380453563295305,
|
| 83 |
+
"rewards/rejected": 8.509196049999446e-05,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.15082956259426847,
|
| 88 |
+
"grad_norm": 297.2610778808594,
|
| 89 |
+
"learning_rate": 4.964280947263677e-06,
|
| 90 |
+
"logits/chosen": -2.456686019897461,
|
| 91 |
+
"logits/rejected": -2.4907171726226807,
|
| 92 |
+
"logps/chosen": -1.7029392719268799,
|
| 93 |
+
"logps/rejected": -1.866715669631958,
|
| 94 |
+
"loss": 2488.6879,
|
| 95 |
+
"rewards/accuracies": 0.6875,
|
| 96 |
+
"rewards/chosen": -0.0010579732479527593,
|
| 97 |
+
"rewards/margins": 0.0010943589732050896,
|
| 98 |
+
"rewards/rejected": -0.0021523318719118834,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.18099547511312217,
|
| 103 |
+
"grad_norm": 418.49908447265625,
|
| 104 |
+
"learning_rate": 4.906048344162677e-06,
|
| 105 |
+
"logits/chosen": -2.526425838470459,
|
| 106 |
+
"logits/rejected": -2.6283812522888184,
|
| 107 |
+
"logps/chosen": -2.2732770442962646,
|
| 108 |
+
"logps/rejected": -2.463239908218384,
|
| 109 |
+
"loss": 2478.1562,
|
| 110 |
+
"rewards/accuracies": 0.625,
|
| 111 |
+
"rewards/chosen": -0.006247864570468664,
|
| 112 |
+
"rewards/margins": 0.0015761550748720765,
|
| 113 |
+
"rewards/rejected": -0.007824019528925419,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.21116138763197587,
|
| 118 |
+
"grad_norm": 884.2985229492188,
|
| 119 |
+
"learning_rate": 4.8209198325401815e-06,
|
| 120 |
+
"logits/chosen": -2.4687628746032715,
|
| 121 |
+
"logits/rejected": -2.5590322017669678,
|
| 122 |
+
"logps/chosen": -3.124673366546631,
|
| 123 |
+
"logps/rejected": -3.5420584678649902,
|
| 124 |
+
"loss": 2466.9961,
|
| 125 |
+
"rewards/accuracies": 0.643750011920929,
|
| 126 |
+
"rewards/chosen": -0.01539889257401228,
|
| 127 |
+
"rewards/margins": 0.003336571855470538,
|
| 128 |
+
"rewards/rejected": -0.018735462799668312,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.24132730015082957,
|
| 133 |
+
"grad_norm": 2458.998291015625,
|
| 134 |
+
"learning_rate": 4.709847017822876e-06,
|
| 135 |
+
"logits/chosen": -2.2200820446014404,
|
| 136 |
+
"logits/rejected": -2.332976818084717,
|
| 137 |
+
"logps/chosen": -5.963655471801758,
|
| 138 |
+
"logps/rejected": -6.942639350891113,
|
| 139 |
+
"loss": 2410.8664,
|
| 140 |
+
"rewards/accuracies": 0.6875,
|
| 141 |
+
"rewards/chosen": -0.04354003071784973,
|
| 142 |
+
"rewards/margins": 0.009163432754576206,
|
| 143 |
+
"rewards/rejected": -0.05270346254110336,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.27149321266968324,
|
| 148 |
+
"grad_norm": 3877.559326171875,
|
| 149 |
+
"learning_rate": 4.5740715227200904e-06,
|
| 150 |
+
"logits/chosen": -2.4160284996032715,
|
| 151 |
+
"logits/rejected": -2.449448585510254,
|
| 152 |
+
"logps/chosen": -12.50175666809082,
|
| 153 |
+
"logps/rejected": -15.007980346679688,
|
| 154 |
+
"loss": 2301.9066,
|
| 155 |
+
"rewards/accuracies": 0.643750011920929,
|
| 156 |
+
"rewards/chosen": -0.10840582847595215,
|
| 157 |
+
"rewards/margins": 0.02461237832903862,
|
| 158 |
+
"rewards/rejected": -0.13301819562911987,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.30165912518853694,
|
| 163 |
+
"grad_norm": 5709.41943359375,
|
| 164 |
+
"learning_rate": 4.415111107797445e-06,
|
| 165 |
+
"logits/chosen": -3.1761019229888916,
|
| 166 |
+
"logits/rejected": -3.347121000289917,
|
| 167 |
+
"logps/chosen": -13.674809455871582,
|
| 168 |
+
"logps/rejected": -16.90283203125,
|
| 169 |
+
"loss": 2178.8832,
|
| 170 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 171 |
+
"rewards/chosen": -0.12066298723220825,
|
| 172 |
+
"rewards/margins": 0.03184697404503822,
|
| 173 |
+
"rewards/rejected": -0.15250995755195618,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.30165912518853694,
|
| 178 |
+
"eval_logits/chosen": -3.1136586666107178,
|
| 179 |
+
"eval_logits/rejected": -3.319279909133911,
|
| 180 |
+
"eval_logps/chosen": -11.729300498962402,
|
| 181 |
+
"eval_logps/rejected": -16.831649780273438,
|
| 182 |
+
"eval_loss": 2139.634033203125,
|
| 183 |
+
"eval_rewards/accuracies": 0.6780821681022644,
|
| 184 |
+
"eval_rewards/chosen": -0.10137370228767395,
|
| 185 |
+
"eval_rewards/margins": 0.050448909401893616,
|
| 186 |
+
"eval_rewards/rejected": -0.15182259678840637,
|
| 187 |
+
"eval_runtime": 118.6764,
|
| 188 |
+
"eval_samples_per_second": 39.233,
|
| 189 |
+
"eval_steps_per_second": 0.615,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.33182503770739064,
|
| 194 |
+
"grad_norm": 6360.60302734375,
|
| 195 |
+
"learning_rate": 4.2347427052552725e-06,
|
| 196 |
+
"logits/chosen": -3.4674181938171387,
|
| 197 |
+
"logits/rejected": -3.7290732860565186,
|
| 198 |
+
"logps/chosen": -14.259218215942383,
|
| 199 |
+
"logps/rejected": -19.902536392211914,
|
| 200 |
+
"loss": 2163.7926,
|
| 201 |
+
"rewards/accuracies": 0.668749988079071,
|
| 202 |
+
"rewards/chosen": -0.12664386630058289,
|
| 203 |
+
"rewards/margins": 0.05585067346692085,
|
| 204 |
+
"rewards/rejected": -0.18249455094337463,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.36199095022624433,
|
| 209 |
+
"grad_norm": 5877.59765625,
|
| 210 |
+
"learning_rate": 4.034982555568005e-06,
|
| 211 |
+
"logits/chosen": -3.6185920238494873,
|
| 212 |
+
"logits/rejected": -3.8847007751464844,
|
| 213 |
+
"logps/chosen": -9.406723976135254,
|
| 214 |
+
"logps/rejected": -15.053186416625977,
|
| 215 |
+
"loss": 2148.0309,
|
| 216 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 217 |
+
"rewards/chosen": -0.07808898389339447,
|
| 218 |
+
"rewards/margins": 0.056064266711473465,
|
| 219 |
+
"rewards/rejected": -0.13415324687957764,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.39215686274509803,
|
| 224 |
+
"grad_norm": 5015.537109375,
|
| 225 |
+
"learning_rate": 3.8180636690262565e-06,
|
| 226 |
+
"logits/chosen": -4.268899440765381,
|
| 227 |
+
"logits/rejected": -4.4745073318481445,
|
| 228 |
+
"logps/chosen": -15.916064262390137,
|
| 229 |
+
"logps/rejected": -19.643577575683594,
|
| 230 |
+
"loss": 2108.7164,
|
| 231 |
+
"rewards/accuracies": 0.581250011920929,
|
| 232 |
+
"rewards/chosen": -0.1432579904794693,
|
| 233 |
+
"rewards/margins": 0.0371844656765461,
|
| 234 |
+
"rewards/rejected": -0.1804424524307251,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.42232277526395173,
|
| 239 |
+
"grad_norm": 6310.57421875,
|
| 240 |
+
"learning_rate": 3.5864108641267815e-06,
|
| 241 |
+
"logits/chosen": -4.173009395599365,
|
| 242 |
+
"logits/rejected": -4.646049976348877,
|
| 243 |
+
"logps/chosen": -11.417980194091797,
|
| 244 |
+
"logps/rejected": -18.31503677368164,
|
| 245 |
+
"loss": 2076.9801,
|
| 246 |
+
"rewards/accuracies": 0.6875,
|
| 247 |
+
"rewards/chosen": -0.09848929196596146,
|
| 248 |
+
"rewards/margins": 0.06831111013889313,
|
| 249 |
+
"rewards/rejected": -0.1668003797531128,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.45248868778280543,
|
| 254 |
+
"grad_norm": 6660.42333984375,
|
| 255 |
+
"learning_rate": 3.3426136618426045e-06,
|
| 256 |
+
"logits/chosen": -4.818399906158447,
|
| 257 |
+
"logits/rejected": -5.288527488708496,
|
| 258 |
+
"logps/chosen": -12.7257080078125,
|
| 259 |
+
"logps/rejected": -20.176372528076172,
|
| 260 |
+
"loss": 2057.0469,
|
| 261 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 262 |
+
"rewards/chosen": -0.11181428283452988,
|
| 263 |
+
"rewards/margins": 0.07331486791372299,
|
| 264 |
+
"rewards/rejected": -0.18512913584709167,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.48265460030165913,
|
| 269 |
+
"grad_norm": 6026.16943359375,
|
| 270 |
+
"learning_rate": 3.089397338773569e-06,
|
| 271 |
+
"logits/chosen": -5.171442985534668,
|
| 272 |
+
"logits/rejected": -5.675396919250488,
|
| 273 |
+
"logps/chosen": -11.334589958190918,
|
| 274 |
+
"logps/rejected": -18.081167221069336,
|
| 275 |
+
"loss": 2062.909,
|
| 276 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 277 |
+
"rewards/chosen": -0.09671419113874435,
|
| 278 |
+
"rewards/margins": 0.06745417416095734,
|
| 279 |
+
"rewards/rejected": -0.1641683727502823,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.5128205128205128,
|
| 284 |
+
"grad_norm": 8983.7353515625,
|
| 285 |
+
"learning_rate": 2.829592462758401e-06,
|
| 286 |
+
"logits/chosen": -6.229965686798096,
|
| 287 |
+
"logits/rejected": -6.906668663024902,
|
| 288 |
+
"logps/chosen": -21.240734100341797,
|
| 289 |
+
"logps/rejected": -27.307592391967773,
|
| 290 |
+
"loss": 2008.2311,
|
| 291 |
+
"rewards/accuracies": 0.65625,
|
| 292 |
+
"rewards/chosen": -0.19601497054100037,
|
| 293 |
+
"rewards/margins": 0.06066788360476494,
|
| 294 |
+
"rewards/rejected": -0.2566828429698944,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.5429864253393665,
|
| 299 |
+
"grad_norm": 6153.8115234375,
|
| 300 |
+
"learning_rate": 2.566103251493184e-06,
|
| 301 |
+
"logits/chosen": -5.295260429382324,
|
| 302 |
+
"logits/rejected": -5.839069366455078,
|
| 303 |
+
"logps/chosen": -13.433932304382324,
|
| 304 |
+
"logps/rejected": -21.858623504638672,
|
| 305 |
+
"loss": 2007.6971,
|
| 306 |
+
"rewards/accuracies": 0.625,
|
| 307 |
+
"rewards/chosen": -0.11825932562351227,
|
| 308 |
+
"rewards/margins": 0.0836234837770462,
|
| 309 |
+
"rewards/rejected": -0.20188280940055847,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5731523378582202,
|
| 314 |
+
"grad_norm": 9230.3251953125,
|
| 315 |
+
"learning_rate": 2.3018751078580287e-06,
|
| 316 |
+
"logits/chosen": -5.80704927444458,
|
| 317 |
+
"logits/rejected": -6.411242485046387,
|
| 318 |
+
"logps/chosen": -17.966712951660156,
|
| 319 |
+
"logps/rejected": -25.16571044921875,
|
| 320 |
+
"loss": 2067.9564,
|
| 321 |
+
"rewards/accuracies": 0.643750011920929,
|
| 322 |
+
"rewards/chosen": -0.1631396859884262,
|
| 323 |
+
"rewards/margins": 0.071785569190979,
|
| 324 |
+
"rewards/rejected": -0.2349252700805664,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6033182503770739,
|
| 329 |
+
"grad_norm": 7809.404296875,
|
| 330 |
+
"learning_rate": 2.0398616948569495e-06,
|
| 331 |
+
"logits/chosen": -6.302067756652832,
|
| 332 |
+
"logits/rejected": -7.327548027038574,
|
| 333 |
+
"logps/chosen": -22.0832576751709,
|
| 334 |
+
"logps/rejected": -31.016321182250977,
|
| 335 |
+
"loss": 1986.1408,
|
| 336 |
+
"rewards/accuracies": 0.65625,
|
| 337 |
+
"rewards/chosen": -0.20454540848731995,
|
| 338 |
+
"rewards/margins": 0.08862989395856857,
|
| 339 |
+
"rewards/rejected": -0.2931753098964691,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6033182503770739,
|
| 344 |
+
"eval_logits/chosen": -5.291009902954102,
|
| 345 |
+
"eval_logits/rejected": -6.192085266113281,
|
| 346 |
+
"eval_logps/chosen": -15.668704986572266,
|
| 347 |
+
"eval_logps/rejected": -24.450298309326172,
|
| 348 |
+
"eval_loss": 1989.8780517578125,
|
| 349 |
+
"eval_rewards/accuracies": 0.7054794430732727,
|
| 350 |
+
"eval_rewards/chosen": -0.14076770842075348,
|
| 351 |
+
"eval_rewards/margins": 0.08724134415388107,
|
| 352 |
+
"eval_rewards/rejected": -0.22800907492637634,
|
| 353 |
+
"eval_runtime": 118.4713,
|
| 354 |
+
"eval_samples_per_second": 39.301,
|
| 355 |
+
"eval_steps_per_second": 0.616,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.6334841628959276,
|
| 360 |
+
"grad_norm": 7728.20361328125,
|
| 361 |
+
"learning_rate": 1.7829919182222752e-06,
|
| 362 |
+
"logits/chosen": -5.712789058685303,
|
| 363 |
+
"logits/rejected": -6.6895551681518555,
|
| 364 |
+
"logps/chosen": -16.13177490234375,
|
| 365 |
+
"logps/rejected": -24.802227020263672,
|
| 366 |
+
"loss": 2035.7111,
|
| 367 |
+
"rewards/accuracies": 0.65625,
|
| 368 |
+
"rewards/chosen": -0.14506369829177856,
|
| 369 |
+
"rewards/margins": 0.08624090254306793,
|
| 370 |
+
"rewards/rejected": -0.2313046157360077,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.6636500754147813,
|
| 375 |
+
"grad_norm": 6645.58544921875,
|
| 376 |
+
"learning_rate": 1.5341371857671782e-06,
|
| 377 |
+
"logits/chosen": -6.007805824279785,
|
| 378 |
+
"logits/rejected": -6.709867000579834,
|
| 379 |
+
"logps/chosen": -17.56117057800293,
|
| 380 |
+
"logps/rejected": -25.152942657470703,
|
| 381 |
+
"loss": 1994.7727,
|
| 382 |
+
"rewards/accuracies": 0.668749988079071,
|
| 383 |
+
"rewards/chosen": -0.15931251645088196,
|
| 384 |
+
"rewards/margins": 0.07542156428098679,
|
| 385 |
+
"rewards/rejected": -0.23473408818244934,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.693815987933635,
|
| 390 |
+
"grad_norm": 9661.85546875,
|
| 391 |
+
"learning_rate": 1.2960793094762347e-06,
|
| 392 |
+
"logits/chosen": -5.852481842041016,
|
| 393 |
+
"logits/rejected": -6.616735935211182,
|
| 394 |
+
"logps/chosen": -17.260818481445312,
|
| 395 |
+
"logps/rejected": -24.995773315429688,
|
| 396 |
+
"loss": 2002.0502,
|
| 397 |
+
"rewards/accuracies": 0.625,
|
| 398 |
+
"rewards/chosen": -0.1562873125076294,
|
| 399 |
+
"rewards/margins": 0.0772886723279953,
|
| 400 |
+
"rewards/rejected": -0.2335759848356247,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.7239819004524887,
|
| 405 |
+
"grad_norm": 6644.81982421875,
|
| 406 |
+
"learning_rate": 1.0714794091391074e-06,
|
| 407 |
+
"logits/chosen": -5.791241645812988,
|
| 408 |
+
"logits/rejected": -6.980679512023926,
|
| 409 |
+
"logps/chosen": -17.29695701599121,
|
| 410 |
+
"logps/rejected": -27.567981719970703,
|
| 411 |
+
"loss": 2000.973,
|
| 412 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 413 |
+
"rewards/chosen": -0.1567213237285614,
|
| 414 |
+
"rewards/margins": 0.1022886261343956,
|
| 415 |
+
"rewards/rejected": -0.2590099275112152,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.7541478129713424,
|
| 420 |
+
"grad_norm": 7918.15234375,
|
| 421 |
+
"learning_rate": 8.628481651367876e-07,
|
| 422 |
+
"logits/chosen": -5.269460201263428,
|
| 423 |
+
"logits/rejected": -6.024519920349121,
|
| 424 |
+
"logps/chosen": -14.571146965026855,
|
| 425 |
+
"logps/rejected": -20.592914581298828,
|
| 426 |
+
"loss": 2004.9424,
|
| 427 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 428 |
+
"rewards/chosen": -0.12985333800315857,
|
| 429 |
+
"rewards/margins": 0.059926945716142654,
|
| 430 |
+
"rewards/rejected": -0.18978026509284973,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.7843137254901961,
|
| 435 |
+
"grad_norm": 11115.212890625,
|
| 436 |
+
"learning_rate": 6.72517752908321e-07,
|
| 437 |
+
"logits/chosen": -5.319769859313965,
|
| 438 |
+
"logits/rejected": -6.440445899963379,
|
| 439 |
+
"logps/chosen": -16.488807678222656,
|
| 440 |
+
"logps/rejected": -25.459197998046875,
|
| 441 |
+
"loss": 1987.9658,
|
| 442 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 443 |
+
"rewards/chosen": -0.14929898083209991,
|
| 444 |
+
"rewards/margins": 0.08880755305290222,
|
| 445 |
+
"rewards/rejected": -0.23810651898384094,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.8144796380090498,
|
| 450 |
+
"grad_norm": 8613.94921875,
|
| 451 |
+
"learning_rate": 5.026157728273967e-07,
|
| 452 |
+
"logits/chosen": -5.407899379730225,
|
| 453 |
+
"logits/rejected": -6.104758262634277,
|
| 454 |
+
"logps/chosen": -14.600807189941406,
|
| 455 |
+
"logps/rejected": -21.75990867614746,
|
| 456 |
+
"loss": 1966.2213,
|
| 457 |
+
"rewards/accuracies": 0.6875,
|
| 458 |
+
"rewards/chosen": -0.13016536831855774,
|
| 459 |
+
"rewards/margins": 0.07122168689966202,
|
| 460 |
+
"rewards/rejected": -0.20138704776763916,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.8446455505279035,
|
| 465 |
+
"grad_norm": 8552.6865234375,
|
| 466 |
+
"learning_rate": 3.5504146691255736e-07,
|
| 467 |
+
"logits/chosen": -5.882079124450684,
|
| 468 |
+
"logits/rejected": -6.893612861633301,
|
| 469 |
+
"logps/chosen": -19.17007064819336,
|
| 470 |
+
"logps/rejected": -29.472118377685547,
|
| 471 |
+
"loss": 1977.0953,
|
| 472 |
+
"rewards/accuracies": 0.699999988079071,
|
| 473 |
+
"rewards/chosen": -0.17585688829421997,
|
| 474 |
+
"rewards/margins": 0.10231776535511017,
|
| 475 |
+
"rewards/rejected": -0.27817463874816895,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.8748114630467572,
|
| 480 |
+
"grad_norm": 13345.8681640625,
|
| 481 |
+
"learning_rate": 2.3144448823151394e-07,
|
| 482 |
+
"logits/chosen": -6.029318332672119,
|
| 483 |
+
"logits/rejected": -6.946678161621094,
|
| 484 |
+
"logps/chosen": -20.550439834594727,
|
| 485 |
+
"logps/rejected": -28.8807373046875,
|
| 486 |
+
"loss": 2005.8379,
|
| 487 |
+
"rewards/accuracies": 0.625,
|
| 488 |
+
"rewards/chosen": -0.18934504687786102,
|
| 489 |
+
"rewards/margins": 0.08309401571750641,
|
| 490 |
+
"rewards/rejected": -0.27243906259536743,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.9049773755656109,
|
| 495 |
+
"grad_norm": 7972.18017578125,
|
| 496 |
+
"learning_rate": 1.3320646032487394e-07,
|
| 497 |
+
"logits/chosen": -5.894993305206299,
|
| 498 |
+
"logits/rejected": -6.6036481857299805,
|
| 499 |
+
"logps/chosen": -18.44249153137207,
|
| 500 |
+
"logps/rejected": -27.600574493408203,
|
| 501 |
+
"loss": 2000.1098,
|
| 502 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 503 |
+
"rewards/chosen": -0.16788385808467865,
|
| 504 |
+
"rewards/margins": 0.0909862369298935,
|
| 505 |
+
"rewards/rejected": -0.25887009501457214,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.9049773755656109,
|
| 510 |
+
"eval_logits/chosen": -5.0283427238464355,
|
| 511 |
+
"eval_logits/rejected": -6.058168411254883,
|
| 512 |
+
"eval_logps/chosen": -15.66148567199707,
|
| 513 |
+
"eval_logps/rejected": -25.235349655151367,
|
| 514 |
+
"eval_loss": 1964.476806640625,
|
| 515 |
+
"eval_rewards/accuracies": 0.7243150472640991,
|
| 516 |
+
"eval_rewards/chosen": -0.14069554209709167,
|
| 517 |
+
"eval_rewards/margins": 0.09516405314207077,
|
| 518 |
+
"eval_rewards/rejected": -0.23585957288742065,
|
| 519 |
+
"eval_runtime": 118.2884,
|
| 520 |
+
"eval_samples_per_second": 39.361,
|
| 521 |
+
"eval_steps_per_second": 0.617,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.9351432880844646,
|
| 526 |
+
"grad_norm": 10022.904296875,
|
| 527 |
+
"learning_rate": 6.142553278648239e-08,
|
| 528 |
+
"logits/chosen": -5.300992012023926,
|
| 529 |
+
"logits/rejected": -6.482455253601074,
|
| 530 |
+
"logps/chosen": -14.489970207214355,
|
| 531 |
+
"logps/rejected": -24.03298568725586,
|
| 532 |
+
"loss": 1974.7594,
|
| 533 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 534 |
+
"rewards/chosen": -0.1287548840045929,
|
| 535 |
+
"rewards/margins": 0.09496350586414337,
|
| 536 |
+
"rewards/rejected": -0.22371840476989746,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.9653092006033183,
|
| 541 |
+
"grad_norm": 10649.0498046875,
|
| 542 |
+
"learning_rate": 1.6904105645142443e-08,
|
| 543 |
+
"logits/chosen": -5.454372406005859,
|
| 544 |
+
"logits/rejected": -6.349032402038574,
|
| 545 |
+
"logps/chosen": -14.140310287475586,
|
| 546 |
+
"logps/rejected": -23.745637893676758,
|
| 547 |
+
"loss": 2000.5619,
|
| 548 |
+
"rewards/accuracies": 0.71875,
|
| 549 |
+
"rewards/chosen": -0.12575171887874603,
|
| 550 |
+
"rewards/margins": 0.09515827149152756,
|
| 551 |
+
"rewards/rejected": -0.220909982919693,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.995475113122172,
|
| 556 |
+
"grad_norm": 9284.1181640625,
|
| 557 |
+
"learning_rate": 1.398597702123583e-10,
|
| 558 |
+
"logits/chosen": -5.438736438751221,
|
| 559 |
+
"logits/rejected": -6.345780849456787,
|
| 560 |
+
"logps/chosen": -16.6071720123291,
|
| 561 |
+
"logps/rejected": -23.371334075927734,
|
| 562 |
+
"loss": 2015.8324,
|
| 563 |
+
"rewards/accuracies": 0.637499988079071,
|
| 564 |
+
"rewards/chosen": -0.15046890079975128,
|
| 565 |
+
"rewards/margins": 0.0670023113489151,
|
| 566 |
+
"rewards/rejected": -0.21747121214866638,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.9984917043740573,
|
| 571 |
+
"step": 331,
|
| 572 |
+
"total_flos": 0.0,
|
| 573 |
+
"train_loss": 2150.5997450906343,
|
| 574 |
+
"train_runtime": 2663.9905,
|
| 575 |
+
"train_samples_per_second": 15.924,
|
| 576 |
+
"train_steps_per_second": 0.124
|
| 577 |
+
}
|
| 578 |
+
],
|
| 579 |
+
"logging_steps": 10,
|
| 580 |
+
"max_steps": 331,
|
| 581 |
+
"num_input_tokens_seen": 0,
|
| 582 |
+
"num_train_epochs": 1,
|
| 583 |
+
"save_steps": 100,
|
| 584 |
+
"stateful_callbacks": {
|
| 585 |
+
"TrainerControl": {
|
| 586 |
+
"args": {
|
| 587 |
+
"should_epoch_stop": false,
|
| 588 |
+
"should_evaluate": false,
|
| 589 |
+
"should_log": false,
|
| 590 |
+
"should_save": true,
|
| 591 |
+
"should_training_stop": true
|
| 592 |
+
},
|
| 593 |
+
"attributes": {}
|
| 594 |
+
}
|
| 595 |
+
},
|
| 596 |
+
"total_flos": 0.0,
|
| 597 |
+
"train_batch_size": 4,
|
| 598 |
+
"trial_name": null,
|
| 599 |
+
"trial_params": null
|
| 600 |
+
}
|