Hugging Face's logo

satpalsr
/
phi2-filter2

Text Generation
Transformers
PyTorch
Safetensors
phi-msft
axolotl
Generated from Trainer
custom_code
Inference Endpoints
Model card Files Community
phi2-filter2 / README.md
satpalsr's picture
satpalsr
End of training
6e609aa verified
preview code
|
raw
history blame
No virus
3.77 kB
metadata
license: mit
base_model: microsoft/phi-2
tags:
  - axolotl
  - generated_from_trainer
model-index:
  - name: phi2-filter2
    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

load_in_8bit: false
load_in_4bit: false
strict: false

hub_model_id: satpalsr/phi2-filter2
hf_use_auth_token: true

datasets:
  - path: satpalsr/translation-filter
    type: completion

dataset_prepared_path:
val_set_size: 0.01
output_dir: ./phi-sft-out2

sequence_len: 2048
sample_packing: false  # currently unsupported
pad_to_sequence_len:

adapter:
lora_model_dir:
lora_r: 16
lora_alpha: 32
lora_dropout: 0.1
lora_target_linear: true
lora_fan_in_fan_out:
lora_modules_to_save:
  - embd
  - lm_head

wandb_project: phi2transfilter
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 16
num_epochs: 4
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: false

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-filter2

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: 0.1944

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: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-05
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 100
  • num_epochs: 4

Training results

Training Loss Epoch Step Validation Loss
2.5676 0.01 1 2.5391
2.4364 0.25 29 2.4042
1.9523 0.5 58 1.8580
1.1137 0.75 87 0.9535
0.5107 1.0 116 0.4195
0.4588 1.25 145 0.2877
0.2876 1.5 174 0.2462
0.2959 1.75 203 0.2264
0.2197 2.0 232 0.2114
0.3045 2.25 261 0.2052
0.2726 2.5 290 0.2022
0.3046 2.75 319 0.1975
0.3316 3.0 348 0.1954
0.2223 3.25 377 0.1950
0.2609 3.5 406 0.1946
0.2739 3.75 435 0.1945
0.2703 4.0 464 0.1944

Framework versions

  • Transformers 4.37.0.dev0
  • Pytorch 2.1.1+cu118
  • Datasets 2.16.1
  • Tokenizers 0.15.0