Upload axolotl_config.yml with huggingface_hub
Browse files- axolotl_config.yml +107 -2
axolotl_config.yml
CHANGED
|
@@ -1,3 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
base_model: llama3-8B
|
| 2 |
model_type: LlamaForCausalLM
|
| 3 |
tokenizer_type: AutoTokenizer
|
|
@@ -53,7 +71,7 @@ group_by_length: false
|
|
| 53 |
bf16: auto
|
| 54 |
fp16:
|
| 55 |
tf32: false
|
| 56 |
-
chat_template:
|
| 57 |
|
| 58 |
gradient_checkpointing: true
|
| 59 |
gradient_checkpointing_kwargs:
|
|
@@ -94,4 +112,91 @@ tokens:
|
|
| 94 |
- "<|im_end|>"
|
| 95 |
lora_modules_to_save:
|
| 96 |
- embed_tokens
|
| 97 |
-
- lm_head
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
base_model: llama3-8B
|
| 6 |
+
model-index:
|
| 7 |
+
- name: qlora_decrease_lr_promptfix
|
| 8 |
+
results: []
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 12 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 13 |
+
|
| 14 |
+
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
|
| 15 |
+
<details><summary>See axolotl config</summary>
|
| 16 |
+
|
| 17 |
+
axolotl version: `0.4.0`
|
| 18 |
+
```yaml
|
| 19 |
base_model: llama3-8B
|
| 20 |
model_type: LlamaForCausalLM
|
| 21 |
tokenizer_type: AutoTokenizer
|
|
|
|
| 71 |
bf16: auto
|
| 72 |
fp16:
|
| 73 |
tf32: false
|
| 74 |
+
chat_template: alpaca
|
| 75 |
|
| 76 |
gradient_checkpointing: true
|
| 77 |
gradient_checkpointing_kwargs:
|
|
|
|
| 112 |
- "<|im_end|>"
|
| 113 |
lora_modules_to_save:
|
| 114 |
- embed_tokens
|
| 115 |
+
- lm_head
|
| 116 |
+
```
|
| 117 |
+
|
| 118 |
+
</details><br>
|
| 119 |
+
|
| 120 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/muellerzr/llama-3-8b-self-align-axolotl/runs/2q8jhm3e)
|
| 121 |
+
# qlora_decrease_lr_promptfix
|
| 122 |
+
|
| 123 |
+
This model was trained from scratch on the None dataset.
|
| 124 |
+
It achieves the following results on the evaluation set:
|
| 125 |
+
- Loss: 0.4121
|
| 126 |
+
|
| 127 |
+
## Model description
|
| 128 |
+
|
| 129 |
+
More information needed
|
| 130 |
+
|
| 131 |
+
## Intended uses & limitations
|
| 132 |
+
|
| 133 |
+
More information needed
|
| 134 |
+
|
| 135 |
+
## Training and evaluation data
|
| 136 |
+
|
| 137 |
+
More information needed
|
| 138 |
+
|
| 139 |
+
## Training procedure
|
| 140 |
+
|
| 141 |
+
### Training hyperparameters
|
| 142 |
+
|
| 143 |
+
The following hyperparameters were used during training:
|
| 144 |
+
- learning_rate: 2e-05
|
| 145 |
+
- train_batch_size: 2
|
| 146 |
+
- eval_batch_size: 2
|
| 147 |
+
- seed: 42
|
| 148 |
+
- distributed_type: multi-GPU
|
| 149 |
+
- num_devices: 2
|
| 150 |
+
- gradient_accumulation_steps: 8
|
| 151 |
+
- total_train_batch_size: 32
|
| 152 |
+
- total_eval_batch_size: 4
|
| 153 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 154 |
+
- lr_scheduler_type: cosine
|
| 155 |
+
- lr_scheduler_warmup_steps: 100
|
| 156 |
+
- num_epochs: 4
|
| 157 |
+
|
| 158 |
+
### Training results
|
| 159 |
+
|
| 160 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
| 161 |
+
|:-------------:|:------:|:----:|:---------------:|
|
| 162 |
+
| 0.6903 | 0.0061 | 1 | 0.6706 |
|
| 163 |
+
| 0.6463 | 0.1285 | 21 | 0.6392 |
|
| 164 |
+
| 0.4944 | 0.2571 | 42 | 0.4806 |
|
| 165 |
+
| 0.4495 | 0.3856 | 63 | 0.4532 |
|
| 166 |
+
| 0.4444 | 0.5142 | 84 | 0.4406 |
|
| 167 |
+
| 0.4185 | 0.6427 | 105 | 0.4334 |
|
| 168 |
+
| 0.4336 | 0.7712 | 126 | 0.4286 |
|
| 169 |
+
| 0.4061 | 0.8998 | 147 | 0.4252 |
|
| 170 |
+
| 0.4002 | 1.0145 | 168 | 0.4221 |
|
| 171 |
+
| 0.4013 | 1.1431 | 189 | 0.4205 |
|
| 172 |
+
| 0.3674 | 1.2716 | 210 | 0.4189 |
|
| 173 |
+
| 0.3942 | 1.4002 | 231 | 0.4175 |
|
| 174 |
+
| 0.3984 | 1.5287 | 252 | 0.4165 |
|
| 175 |
+
| 0.3867 | 1.6572 | 273 | 0.4150 |
|
| 176 |
+
| 0.3872 | 1.7858 | 294 | 0.4137 |
|
| 177 |
+
| 0.401 | 1.9143 | 315 | 0.4130 |
|
| 178 |
+
| 0.3602 | 2.0275 | 336 | 0.4126 |
|
| 179 |
+
| 0.3817 | 2.1561 | 357 | 0.4131 |
|
| 180 |
+
| 0.3592 | 2.2846 | 378 | 0.4129 |
|
| 181 |
+
| 0.3729 | 2.4132 | 399 | 0.4127 |
|
| 182 |
+
| 0.372 | 2.5417 | 420 | 0.4121 |
|
| 183 |
+
| 0.3685 | 2.6702 | 441 | 0.4120 |
|
| 184 |
+
| 0.3732 | 2.7988 | 462 | 0.4115 |
|
| 185 |
+
| 0.38 | 2.9273 | 483 | 0.4112 |
|
| 186 |
+
| 0.3637 | 3.0413 | 504 | 0.4114 |
|
| 187 |
+
| 0.3628 | 3.1699 | 525 | 0.4118 |
|
| 188 |
+
| 0.355 | 3.2984 | 546 | 0.4122 |
|
| 189 |
+
| 0.3646 | 3.4269 | 567 | 0.4121 |
|
| 190 |
+
| 0.3496 | 3.5555 | 588 | 0.4121 |
|
| 191 |
+
| 0.3573 | 3.6840 | 609 | 0.4121 |
|
| 192 |
+
| 0.3598 | 3.8125 | 630 | 0.4121 |
|
| 193 |
+
| 0.3669 | 3.9411 | 651 | 0.4121 |
|
| 194 |
+
|
| 195 |
+
|
| 196 |
+
### Framework versions
|
| 197 |
+
|
| 198 |
+
- PEFT 0.11.1
|
| 199 |
+
- Transformers 4.42.0.dev0
|
| 200 |
+
- Pytorch 2.3.0+cu118
|
| 201 |
+
- Datasets 2.19.1
|
| 202 |
+
- Tokenizers 0.19.1
|