File size: 7,609 Bytes
210bcff ca51f40 210bcff ca51f40 210bcff ca51f40 210bcff 9bbfa57 210bcff ca51f40 210bcff |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 |
---
license: llama2
library_name: peft
tags:
- trl
- dpo
- generated_from_trainer
base_model: meta-llama/Llama-2-7b-hf
model-index:
- name: llama_DPO_model_e2
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# llama_DPO_model_e2
This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1001
- Rewards/chosen: 0.4226
- Rewards/rejected: -1.9804
- Rewards/accuracies: 1.0
- Rewards/margins: 2.4030
- Logps/rejected: -204.6132
- Logps/chosen: -156.4080
- Logits/rejected: -1.0519
- Logits/chosen: -0.8585
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 6e-07
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
| 0.6757 | 0.1 | 25 | 0.6650 | 0.0149 | -0.0435 | 0.7767 | 0.0584 | -185.2444 | -160.4850 | -1.0519 | -0.8543 |
| 0.6136 | 0.2 | 50 | 0.5989 | 0.0552 | -0.1462 | 0.9567 | 0.2014 | -186.2718 | -160.0822 | -1.0523 | -0.8553 |
| 0.5526 | 0.3 | 75 | 0.5225 | 0.1032 | -0.2804 | 1.0 | 0.3837 | -187.6138 | -159.6014 | -1.0520 | -0.8542 |
| 0.4819 | 0.4 | 100 | 0.4502 | 0.1474 | -0.4325 | 0.9967 | 0.5798 | -189.1341 | -159.1602 | -1.0518 | -0.8548 |
| 0.4253 | 0.5 | 125 | 0.3835 | 0.1905 | -0.5943 | 1.0 | 0.7848 | -190.7523 | -158.7284 | -1.0527 | -0.8564 |
| 0.3448 | 0.6 | 150 | 0.3197 | 0.2328 | -0.7813 | 1.0 | 1.0141 | -192.6229 | -158.3063 | -1.0526 | -0.8559 |
| 0.3007 | 0.7 | 175 | 0.2637 | 0.2788 | -0.9753 | 1.0 | 1.2542 | -194.5630 | -157.8456 | -1.0525 | -0.8586 |
| 0.2369 | 0.79 | 200 | 0.2192 | 0.3135 | -1.1671 | 1.0 | 1.4807 | -196.4808 | -157.4985 | -1.0519 | -0.8604 |
| 0.1987 | 0.89 | 225 | 0.1825 | 0.3436 | -1.3550 | 1.0 | 1.6986 | -198.3592 | -157.1976 | -1.0520 | -0.8594 |
| 0.1616 | 0.99 | 250 | 0.1532 | 0.3687 | -1.5379 | 1.0 | 1.9066 | -200.1886 | -156.9470 | -1.0519 | -0.8604 |
| 0.1525 | 1.09 | 275 | 0.1346 | 0.3861 | -1.6703 | 1.0 | 2.0564 | -201.5127 | -156.7730 | -1.0511 | -0.8582 |
| 0.1194 | 1.19 | 300 | 0.1246 | 0.3970 | -1.7483 | 1.0 | 2.1453 | -202.2923 | -156.6637 | -1.0509 | -0.8584 |
| 0.1128 | 1.29 | 325 | 0.1161 | 0.4062 | -1.8227 | 1.0 | 2.2289 | -203.0370 | -156.5718 | -1.0511 | -0.8577 |
| 0.1194 | 1.39 | 350 | 0.1108 | 0.4127 | -1.8680 | 1.0 | 2.2807 | -203.4899 | -156.5069 | -1.0514 | -0.8602 |
| 0.1123 | 1.49 | 375 | 0.1070 | 0.4151 | -1.9092 | 1.0 | 2.3243 | -203.9014 | -156.4828 | -1.0515 | -0.8584 |
| 0.1008 | 1.59 | 400 | 0.1046 | 0.4209 | -1.9290 | 1.0 | 2.3499 | -204.0999 | -156.4248 | -1.0516 | -0.8618 |
| 0.0971 | 1.69 | 425 | 0.1033 | 0.4208 | -1.9461 | 1.0 | 2.3669 | -204.2709 | -156.4260 | -1.0510 | -0.8586 |
| 0.109 | 1.79 | 450 | 0.1019 | 0.4235 | -1.9597 | 1.0 | 2.3832 | -204.4061 | -156.3985 | -1.0510 | -0.8587 |
| 0.1035 | 1.89 | 475 | 0.1009 | 0.4234 | -1.9700 | 1.0 | 2.3934 | -204.5094 | -156.4001 | -1.0517 | -0.8580 |
| 0.1046 | 1.99 | 500 | 0.1004 | 0.4210 | -1.9772 | 1.0 | 2.3983 | -204.5820 | -156.4234 | -1.0511 | -0.8603 |
| 0.0961 | 2.09 | 525 | 0.1002 | 0.4227 | -1.9798 | 1.0 | 2.4025 | -204.6080 | -156.4070 | -1.0518 | -0.8587 |
| 0.0932 | 2.19 | 550 | 0.1000 | 0.4237 | -1.9796 | 1.0 | 2.4033 | -204.6052 | -156.3964 | -1.0518 | -0.8597 |
| 0.0901 | 2.29 | 575 | 0.1002 | 0.4231 | -1.9785 | 1.0 | 2.4015 | -204.5942 | -156.4030 | -1.0514 | -0.8594 |
| 0.1033 | 2.38 | 600 | 0.1003 | 0.4248 | -1.9780 | 1.0 | 2.4028 | -204.5901 | -156.3859 | -1.0517 | -0.8616 |
| 0.1108 | 2.48 | 625 | 0.0999 | 0.4262 | -1.9796 | 1.0 | 2.4057 | -204.6053 | -156.3723 | -1.0517 | -0.8583 |
| 0.1026 | 2.58 | 650 | 0.0998 | 0.4208 | -1.9879 | 1.0 | 2.4088 | -204.6889 | -156.4255 | -1.0522 | -0.8594 |
| 0.0956 | 2.68 | 675 | 0.1001 | 0.4227 | -1.9818 | 1.0 | 2.4045 | -204.6279 | -156.4070 | -1.0517 | -0.8588 |
| 0.1003 | 2.78 | 700 | 0.0996 | 0.4241 | -1.9817 | 1.0 | 2.4058 | -204.6262 | -156.3926 | -1.0516 | -0.8584 |
| 0.0874 | 2.88 | 725 | 0.0997 | 0.4228 | -1.9835 | 1.0 | 2.4064 | -204.6450 | -156.4057 | -1.0519 | -0.8609 |
| 0.1001 | 2.98 | 750 | 0.1001 | 0.4226 | -1.9804 | 1.0 | 2.4030 | -204.6132 | -156.4080 | -1.0519 | -0.8585 |
### Framework versions
- PEFT 0.8.2
- Transformers 4.38.1
- Pytorch 2.2.0+cu118
- Datasets 2.17.1
- Tokenizers 0.15.2 |