Hugging Face's logo

noeloco
/
camel-lora

PEFT
PyTorch
Safetensors
llama
axolotl
Generated from Trainer
8-bit precision
bitsandbytes
Model card Files Community
camel-lora / README.md
noeloco's picture
noeloco
End of training
08b6745 verified
preview code
|
raw
history blame
3.66 kB
metadata
license: llama2
library_name: peft
tags:
  - axolotl
  - generated_from_trainer
base_model: codellama/CodeLlama-7b-hf
model-index:
  - name: camel-lora
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.0

base_model: codellama/CodeLlama-7b-hf
model_type: LlamaForCausalLM
tokenizer_type: CodeLlamaTokenizer
is_llama_derived_model: true

hub_model_id: noeloco/camel-lora

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: noeloco/fizzbuzz-sft
    type: alpaca
    ds_type: json

hf_use_auth_token: true
push_dataset_to_hub: noeloco
val_set_size: 0.05
output_dir: ./lora-out
chat_template: chatml

sequence_len: 4096
sample_packing: false
pad_to_sequence_len: true

adapter: qlora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project: runpod1
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: false
tf32: true

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: 4
saves_per_epoch: 1
debug: true
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
  unk_token: "<unk>"

camel-lora

This model is a fine-tuned version of codellama/CodeLlama-7b-hf on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0290

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
1.7685 0.06 1 2.5524
1.8762 0.29 5 2.4927
1.215 0.57 10 1.4546
0.484 0.86 15 0.7250
0.3667 1.14 20 0.4146
0.1638 1.43 25 0.2123
0.2948 1.71 30 0.0980
0.2003 2.0 35 0.0629
0.0888 2.29 40 0.0577
0.0918 2.57 45 0.0414
0.0931 2.86 50 0.0363
0.0982 3.14 55 0.0304
0.0849 3.43 60 0.0289
0.0511 3.71 65 0.0290

Framework versions

  • PEFT 0.10.1.dev0
  • Transformers 4.40.0.dev0
  • Pytorch 2.1.2+cu118
  • Datasets 2.15.0
  • Tokenizers 0.15.0