[
](https://github.com/axolotl-ai-cloud/axolotl)
See axolotl config
axolotl version: `0.9.1.post1`
```yaml
base_model: meta-llama/Llama-3.1-8B-Instruct
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer
gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 1
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0001
load_in_8bit: true
load_in_4bit: false
adapter: lora
lora_model_dir: null
lora_r: 8
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
- q_proj
- v_proj
- k_proj
datasets:
- path: /workspace/FinLoRA/data/train/xbrl_extract_train.jsonl
type:
system_prompt: ''
field_system: system
field_instruction: context
field_output: target
format: '[INST] {instruction} [/INST]'
no_input_format: '[INST] {instruction} [/INST]'
dataset_prepared_path: null
val_set_size: 0.02
output_dir: /workspace/FinLoRA/lora/axolotl-output/xbrl_extract_llama_3_1_8b_8bits_r8_rslora
peft_use_dora: false
peft_use_rslora: true
sequence_len: 4096
sample_packing: false
pad_to_sequence_len: false
wandb_project: finlora_models
wandb_entity: null
wandb_watch: gradients
wandb_name: xbrl_extract_llama_3_1_8b_8bits_r8_rslora
wandb_log_model: 'false'
bf16: auto
tf32: false
gradient_checkpointing: true
resume_from_checkpoint: null
logging_steps: 500
flash_attention: false
deepspeed: deepspeed_configs/zero1.json
warmup_steps: 10
evals_per_epoch: 4
saves_per_epoch: 1
weight_decay: 0.0
special_tokens:
pad_token: <|end_of_text|>
chat_template: llama3
```
# workspace/FinLoRA/lora/axolotl-output/xbrl_extract_llama_3_1_8b_8bits_r8_rslora
This model is a fine-tuned version of [meta-llama/Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct) on the /workspace/FinLoRA/data/train/xbrl_extract_train.jsonl dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0001
## 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: 0.0001
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- total_eval_batch_size: 4
- optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 1.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:------:|:----:|:---------------:|
| No log | 0.0038 | 1 | 1.6299 |
| No log | 0.2526 | 67 | 0.0012 |
| No log | 0.5052 | 134 | 0.0005 |
| No log | 0.7578 | 201 | 0.0001 |
### Framework versions
- PEFT 0.15.2
- Transformers 4.51.3
- Pytorch 2.8.0.dev20250319+cu128
- Datasets 3.5.1
- Tokenizers 0.21.1