Hugging Face's logo

openaccess-ai-collective
/
phi2-alpaca

Text Generation
Transformers
PyTorch
TensorBoard
Safetensors
phi-msft
axolotl
Generated from Trainer
custom_code
Inference Endpoints
Model card Files Metrics Community
phi2-alpaca / README.md
winglian's picture
winglian
End of training
351f3c3
preview code
|
raw
history blame
2.97 kB
metadata
license: mit
base_model: microsoft/phi-2
tags:
  - axolotl
  - generated_from_trainer
model-index:
  - name: phi2-alpaca
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.3.0

base_model: microsoft/phi-2
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer
trust_remote_code: true

hub_model_id: openaccess-ai-collective/phi2-alpaca

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: tatsu-lab/alpaca
    type: alpaca

dataset_prepared_path:
val_set_size: 0.05
output_dir: ./phi-sft-out

sequence_len: 2048
sample_packing: false  # currently unsupported
pad_to_sequence_len:

wandb_project: phi2
wandb_entity: oaaic
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 4
num_epochs: 1
optimizer: paged_adamw_8bit
adam_beta2: 0.95
adam_epsilon: 0.00001
max_grad_norm: 1.0
lr_scheduler: cosine
learning_rate: 1e-5

train_on_inputs: false
group_by_length: false
bf16: true
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: 100
evals_per_epoch: 4
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.1
fsdp:
fsdp_config:
resize_token_embeddings_to_32x: true
special_tokens:
  pad_token: "<|endoftext|>"

phi2-alpaca

This model is a fine-tuned version of microsoft/phi-2 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.9343

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: 4
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-05
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 1

Training results

Training Loss Epoch Step Validation Loss
1.3994 0.0 1 1.3199
0.9532 0.25 386 0.9886
0.8445 0.5 772 0.9421
0.7303 0.75 1158 0.9343

Framework versions

  • Transformers 4.37.0.dev0
  • Pytorch 2.0.1+cu118
  • Datasets 2.16.1
  • Tokenizers 0.15.0