Hugging Face's logo

QuantFactory
/
Llama-3-8B-WizardLM-196K-GGUF

Text Generation
GGUF
axolotl
Generated from Trainer
Inference Endpoints
conversational
Model card Files Community
Llama-3-8B-WizardLM-196K-GGUF / README.md
munish0838's picture
munish0838
Create README.md
765447a verified
preview code
|
raw
history blame contribute delete
3.18 kB
metadata
license: llama3
base_model: Magpie-Align/Llama-3-8B-WizardLM-196K
tags:
  - axolotl
  - generated_from_trainer
model-index:
  - name: Llama-3-8B-WizardLM-196K
    results: []
pipeline_tag: text-generation

QuantFactory/Llama-3-8B-WizardLM-196K-GGUF

This is quantized version of Magpie-Align/Llama-3-8B-WizardLM-196K created using llama.cpp

Model Description

Built with Axolotl

See axolotl config

axolotl version: 0.4.0

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

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: Leon-Leee/Wizardlm_Evol_Instruct_v2_196K_backuped
    type: sharegpt
    conversation: llama3
dataset_prepared_path: last_run_prepared
val_set_size: 0.001
output_dir: ./out_Llama-8B-WizardLM-196k

sequence_len: 8192
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

wandb_project: SynDa
wandb_entity:
wandb_watch:
wandb_name: Llama-3-8B-WizardLM-196k
wandb_log_model:
hub_model_id: SynDa/Llama-3-8B-WizardLM-196K

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 2
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 2e-5

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 100
evals_per_epoch: 3
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>

Llama-3-8B-WizardLM-196K

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

  • Loss: 0.6077

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 2e-05
  • 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: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss
0.7323 0.0036 1 1.0826
0.5934 0.3344 93 0.6450
0.5497 0.6688 186 0.6192
0.5295 1.0031 279 0.6059
0.4664 1.3236 372 0.6103
0.4729 1.6580 465 0.6077

Framework versions

  • Transformers 4.40.2
  • Pytorch 2.3.0+cu121
  • Datasets 2.19.1
  • Tokenizers 0.19.1