Hugging Face's logo

anastas5
/
llama3.1-8B-Instruct-rus-test-v3

Transformers
PyTorch
Generated from Trainer
Inference Endpoints
Model card Files Community
llama3.1-8B-Instruct-rus-test-v3 / README.md
anastas5's picture
anastas5
Добавлены файлы модели
b535eba
preview code
|
raw
history blame contribute delete
3.86 kB
metadata
library_name: transformers
license: llama3.1
base_model: anastas5/llama3.1-8B-Instruct-rus-test-v2
tags:
  - generated_from_trainer
model-index:
  - name: >-
      home/ubuntu/llm_training/axolotl/llama3-8b-gpt-4o-ru/output_llama3_8b_gpt_4o_ru
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.1

base_model: anastas5/llama3.1-8B-Instruct-rus-test-v2

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: anastas5/dataset-rus-test-six
    type: sharegpt
    conversation: llama-3
dataset_prepared_path: /home/ubuntu/llm_training/axolotl/llama3-8b-gpt-4o-ru/prepared_tagengo_rus
val_set_size: 0.05
output_dir: /home/ubuntu/llm_training/axolotl/llama3-8b-gpt-4o-ru/output_llama3_8b_gpt_4o_ru

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

use_wandb: false

gradient_accumulation_steps: 2
micro_batch_size: 2
num_epochs: 4
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 1e-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: 10
evals_per_epoch: 5
eval_table_size:
saves_per_epoch: 1
debug:
weight_decay: 0.0
special_tokens:
  pad_token: <|end_of_text|>

home/ubuntu/llm_training/axolotl/llama3-8b-gpt-4o-ru/output_llama3_8b_gpt_4o_ru

This model is a fine-tuned version of anastas5/llama3.1-8B-Instruct-rus-test-v2 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7729

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: 1e-05
  • train_batch_size: 2
  • eval_batch_size: 2
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 4
  • 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
0.957 0.0769 1 0.8838
0.9398 0.2308 3 0.8762
1.0969 0.4615 6 0.8373
0.9608 0.6923 9 0.8226
0.8364 0.9231 12 0.8146
0.7566 1.1154 15 0.7914
0.7927 1.3462 18 0.7818
0.74 1.5769 21 0.7784
0.7247 1.8077 24 0.7783
0.7261 2.0385 27 0.7748
0.7255 2.2308 30 0.7727
0.6439 2.4615 33 0.7726
0.5354 2.6923 36 0.7725
0.638 2.9231 39 0.7734
0.6246 3.0769 42 0.7726
0.5374 3.3077 45 0.7726
0.5539 3.5385 48 0.7729
0.6056 3.7692 51 0.7729

Framework versions

  • Transformers 4.45.1
  • Pytorch 2.3.1+cu121
  • Datasets 3.0.1
  • Tokenizers 0.20.1