Magpie-Pro
Collection
Dataset built with Meta Llama 3 70B. Models are fine-tuned from Llama 3 8B.
•
8 items
•
Updated
•
14
Project Web: https://magpie-align.github.io/
Arxiv Technical Report: https://arxiv.org/abs/2406.08464
Codes: https://github.com/magpie-align/magpie
This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B on First 200K data of Magpie-Align/Magpie-Pro-300K-Filtered dataset.
Please use Magpie-Align/Llama-3-8B-Magpie-Pro-SFT-v0.1 with better performance.
The following hyperparameters were used during training:
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.8686 | 0.0018 | 1 | 0.8670 |
0.514 | 0.3342 | 184 | 0.5190 |
0.4769 | 0.6685 | 368 | 0.4684 |
0.4394 | 1.0027 | 552 | 0.4440 |
0.3399 | 1.3224 | 736 | 0.4436 |
0.3394 | 1.6567 | 920 | 0.4413 |
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: Magpie-Align/Magpie-Pro-300K-Filtered-First200K
type: sharegpt
conversation: llama3
dataset_prepared_path: last_run_prepared
val_set_size: 0.001
output_dir: ./out_Llama-3-8B-Magpie-Pro-200K-FilteredL
sequence_len: 8192
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true
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|>