End of training
Browse files- README.md +75 -0
- adapter_model.bin +1 -1
README.md
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: dhmeltzer/llama-7b-SFT_eli5_wiki65k_1024_r_64_alpha_16_merged
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
model-index:
|
| 6 |
+
- name: llama-7b-SFT-qlora-eli5-wiki_DPO_ds_RM_random_1024_r_64_alpha_16
|
| 7 |
+
results: []
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 11 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 12 |
+
|
| 13 |
+
# llama-7b-SFT-qlora-eli5-wiki_DPO_ds_RM_random_1024_r_64_alpha_16
|
| 14 |
+
|
| 15 |
+
This model is a fine-tuned version of [dhmeltzer/llama-7b-SFT_eli5_wiki65k_1024_r_64_alpha_16_merged](https://huggingface.co/dhmeltzer/llama-7b-SFT_eli5_wiki65k_1024_r_64_alpha_16_merged) on an unknown dataset.
|
| 16 |
+
It achieves the following results on the evaluation set:
|
| 17 |
+
- Loss: 0.6799
|
| 18 |
+
- Rewards/chosen: -0.2800
|
| 19 |
+
- Rewards/rejected: -0.3342
|
| 20 |
+
- Rewards/accuracies: 0.5614
|
| 21 |
+
- Rewards/margins: 0.0542
|
| 22 |
+
- Logps/rejected: -203.9819
|
| 23 |
+
- Logps/chosen: -201.2875
|
| 24 |
+
- Logits/rejected: 1.0308
|
| 25 |
+
- Logits/chosen: 1.0515
|
| 26 |
+
|
| 27 |
+
## Model description
|
| 28 |
+
|
| 29 |
+
More information needed
|
| 30 |
+
|
| 31 |
+
## Intended uses & limitations
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Training and evaluation data
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training procedure
|
| 40 |
+
|
| 41 |
+
### Training hyperparameters
|
| 42 |
+
|
| 43 |
+
The following hyperparameters were used during training:
|
| 44 |
+
- learning_rate: 5e-05
|
| 45 |
+
- train_batch_size: 32
|
| 46 |
+
- eval_batch_size: 32
|
| 47 |
+
- seed: 42
|
| 48 |
+
- gradient_accumulation_steps: 4
|
| 49 |
+
- total_train_batch_size: 128
|
| 50 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 51 |
+
- lr_scheduler_type: cosine
|
| 52 |
+
- lr_scheduler_warmup_ratio: 0.03
|
| 53 |
+
- num_epochs: 1
|
| 54 |
+
|
| 55 |
+
### Training results
|
| 56 |
+
|
| 57 |
+
| Training Loss | Epoch | Step | Logits/chosen | Logits/rejected | Logps/chosen | Logps/rejected | Validation Loss | Rewards/accuracies | Rewards/chosen | Rewards/margins | Rewards/rejected |
|
| 58 |
+
|:-------------:|:-----:|:----:|:-------------:|:---------------:|:------------:|:--------------:|:---------------:|:------------------:|:--------------:|:---------------:|:----------------:|
|
| 59 |
+
| 0.6934 | 0.1 | 19 | 1.0457 | 1.0239 | -198.8585 | -201.0544 | 0.6914 | 0.5402 | -0.0371 | 0.0043 | -0.0414 |
|
| 60 |
+
| 0.6881 | 0.21 | 38 | 1.0438 | 1.0224 | -200.6890 | -203.1315 | 0.6838 | 0.5625 | -0.2202 | 0.0290 | -0.2492 |
|
| 61 |
+
| 0.6812 | 0.31 | 57 | 1.0515 | 1.0308 | -201.2875 | -203.9819 | 0.6799 | 0.5614 | -0.2800 | 0.0542 | -0.3342 |
|
| 62 |
+
| 0.6732 | 0.41 | 76 | 1.0596 | 1.0387 | -200.9070 | -203.6835 | 0.6814 | 0.5513 | -0.2420 | 0.0624 | -0.3044 |
|
| 63 |
+
| 0.6613 | 0.52 | 95 | 0.6802 | -0.2778 | -0.3555 | 0.5647 | 0.0777 | -204.1952 | -201.2652 | 1.0369 | 1.0572 |
|
| 64 |
+
| 0.6393 | 0.62 | 114 | 0.6807 | -0.2791 | -0.3688 | 0.5681 | 0.0897 | -204.3279 | -201.2781 | 1.0358 | 1.0562 |
|
| 65 |
+
| 0.6611 | 0.72 | 133 | 0.6841 | -0.2978 | -0.3911 | 0.5692 | 0.0933 | -204.5510 | -201.4653 | 1.0325 | 1.0526 |
|
| 66 |
+
| 0.6414 | 0.83 | 152 | 0.6824 | -0.2883 | -0.3814 | 0.5625 | 0.0930 | -204.4536 | -201.3704 | 1.0310 | 1.0515 |
|
| 67 |
+
| 0.6529 | 0.93 | 171 | 0.6823 | -0.2824 | -0.3743 | 0.5525 | 0.0920 | -204.3832 | -201.3109 | 1.0318 | 1.0522 |
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
### Framework versions
|
| 71 |
+
|
| 72 |
+
- Transformers 4.32.1
|
| 73 |
+
- Pytorch 2.0.1+cu118
|
| 74 |
+
- Datasets 2.14.4
|
| 75 |
+
- Tokenizers 0.13.3
|
adapter_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 639792909
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:465fe924ccf44e7ef6df7fcc5ec700fd43915d1d19f1a26c97028b061b28f543
|
| 3 |
size 639792909
|