jfrickradiant/llama-8b-ft-for-chatbot-example

See axolotl config

axolotl version: 0.4.1

base_model: meta-llama/Meta-Llama-3-8B-Instruct
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer  # PreTrainedTokenizerFast

load_in_8bit: false
load_in_4bit: true 
strict: false

datasets:
  - path: /test-file-system/axolotl/test-file-system/axolotl/ft_data_sharegpt.jsonl 
    type: sharegpt
    conversation: chatml 
    #field_human: user
    #field_model: assistant
    #roles:
    #   input:
    #    - user
    #    - system
    #  output:
    #    - assistant
dataset_prepared_path:
val_set_size: 0.05
output_dir: /test-file-system/axolotl/test-file-system/axolotl/lora-llama3-8b-chat

adapter: qlora
lora_model_dir:

sequence_len: 4096
sample_packing: false 
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: 2
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
s2_attention:

warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
   pad_token: <|end_of_text|>

test-file-system/axolotl/test-file-system/axolotl/lora-llama3-8b-chat

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0002

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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
1.8618	0.0280	1	1.8569
0.0185	0.2517	9	0.0596
0.0056	0.5035	18	0.0202
0.0008	0.7552	27	0.0005
0.0006	1.0070	36	0.0002
0.0001	1.2587	45	0.0000
0.0004	1.5105	54	0.0004
0.0007	1.7622	63	0.0002
0.0001	2.0140	72	0.0001
0.0001	2.2657	81	0.0002
0.0006	2.5175	90	0.0004
0.0006	2.7692	99	0.0004
0.0005	3.0210	108	0.0003
0.0003	3.2727	117	0.0002
0.0004	3.5245	126	0.0002
0.0006	3.7762	135	0.0002

Framework versions

• PEFT 0.11.1
• Transformers 4.41.1
• Pytorch 2.1.2+cu118
• Datasets 2.19.1
• Tokenizers 0.19.1