metadata
library_name: peft
tags:
- generated_from_trainer
base_model: NousResearch/Llama-2-7b-chat-hf
model-index:
- name: qlora-out_7b_chat_llama
results: []
See axolotl config
axolotl version: 0.4.0
# base_model: NousResearch/Llama-2-7b-hf
base_model: NousResearch/Llama-2-7b-chat-hf
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
is_llama_derived_model: true
load_in_8bit: false
load_in_4bit: true
strict: false
datasets:
- path: /home/ubuntu/Project_Files/Finetune/Data/json_files/combined_sentences.json
type: completion
ds_type: json
dataset_prepared_path: last_run_prepared
val_set_size: 0.05
output_dir: ./qlora-out_7b_chat_llama
adapter: qlora
lora_model_dir:
sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:
wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:
gradient_accumulation_steps: 4
micro_batch_size: 4
num_epochs: 4
optimizer: paged_adamw_32bit
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 10
evals_per_epoch: 10
eval_table_size:
saves_per_epoch: 2
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
qlora-out_7b_chat_llama
This model is a fine-tuned version of NousResearch/Llama-2-7b-chat-hf on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5727
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: 0.0002
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 4.1915 | 0.0 | 1 | 4.1955 |
| 0.6659 | 0.1 | 95 | 0.5660 |
| 0.6226 | 0.2 | 190 | 0.5444 |
| 0.6062 | 0.3 | 285 | 0.5330 |
| 0.594 | 0.4 | 380 | 0.5310 |
| 0.5866 | 0.5 | 475 | 0.5322 |
| 0.5801 | 0.6 | 570 | 0.5280 |
| 0.5772 | 0.7 | 665 | 0.5270 |
| 0.5741 | 0.8 | 760 | 0.5285 |
| 0.5719 | 0.9 | 855 | 0.5279 |
| 0.5674 | 1.0 | 950 | 0.5313 |
| 0.5673 | 1.09 | 1045 | 0.5360 |
| 0.5652 | 1.19 | 1140 | 0.5323 |
| 0.5609 | 1.29 | 1235 | 0.5292 |
| 0.5615 | 1.39 | 1330 | 0.5329 |
| 0.5589 | 1.49 | 1425 | 0.5346 |
| 0.5572 | 1.59 | 1520 | 0.5364 |
| 0.5567 | 1.69 | 1615 | 0.5392 |
| 0.557 | 1.8 | 1710 | 0.5396 |
| 0.5549 | 1.9 | 1805 | 0.5454 |
| 0.5517 | 2.0 | 1900 | 0.5475 |
| 0.5493 | 2.08 | 1995 | 0.5515 |
| 0.5506 | 2.18 | 2090 | 0.5544 |
| 0.5497 | 2.29 | 2185 | 0.5507 |
| 0.548 | 2.39 | 2280 | 0.5563 |
| 0.5483 | 2.49 | 2375 | 0.5578 |
| 0.5502 | 2.59 | 2470 | 0.5602 |
| 0.5472 | 2.69 | 2565 | 0.5632 |
| 0.548 | 2.79 | 2660 | 0.5649 |
| 0.5478 | 2.89 | 2755 | 0.5630 |
| 0.5441 | 2.99 | 2850 | 0.5663 |
| 0.5418 | 3.08 | 2945 | 0.5705 |
| 0.5453 | 3.18 | 3040 | 0.5679 |
| 0.5431 | 3.28 | 3135 | 0.5717 |
| 0.5451 | 3.38 | 3230 | 0.5734 |
| 0.5469 | 3.48 | 3325 | 0.5711 |
| 0.5445 | 3.58 | 3420 | 0.5734 |
| 0.5428 | 3.68 | 3515 | 0.5724 |
| 0.5436 | 3.78 | 3610 | 0.5730 |
| 0.546 | 3.88 | 3705 | 0.5727 |
Framework versions
- PEFT 0.8.2
- Transformers 4.38.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0