Hugging Face's logo

FourOhFour
/
Crispy_Crab_4B

Text Generation
Transformers
Safetensors
llama
axolotl
Generated from Trainer
text-generation-inference
Inference Endpoints
Model card Files Community
Crispy_Crab_4B / README.md
jeiku's picture
jeiku
Model save
e03fce1 verified
preview code
|
raw
history blame
3.64 kB
metadata
library_name: transformers
license: other
base_model: jeiku/instructered4B
tags:
  - axolotl
  - generated_from_trainer
model-index:
  - name: TheBest4B
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.1

base_model: jeiku/instructered4B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

hub_model_id: jeiku/TheBest4B
hub_strategy: "all_checkpoints"
push_dataset_to_hub:
hf_use_auth_token: true

datasets:
  - path: FourOhFour/RP_Phase
    type: sharegpt
    conversation: chatml

chat_template: chatml

shuffle_merged_datasets: true
val_set_size: 0.0025
output_dir: ./outputs/out

adapter:
lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:

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

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

wandb_project: EXP4B
wandb_entity:
wandb_watch:
wandb_name: EXP4B
wandb_log_model:

gradient_accumulation_steps: 12
micro_batch_size: 3
num_epochs: 2
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.00001
weight_decay: 0.05

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

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint: 
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_ratio: 0.1
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 2

debug:
deepspeed: deepspeed_configs/zero3_bf16.json
fsdp:
fsdp_config:

special_tokens:
  pad_token: <|finetune_right_pad_id|>

TheBest4B

This model is a fine-tuned version of jeiku/instructered4B on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 2.1148

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: 3
  • eval_batch_size: 3
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 2
  • gradient_accumulation_steps: 12
  • total_train_batch_size: 72
  • total_eval_batch_size: 6
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 22
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss
1.8805 0.0089 1 2.7425
1.7985 0.2491 28 2.2908
1.727 0.4981 56 2.1943
1.7429 0.7472 84 2.1665
1.6867 0.9963 112 2.1309
1.6463 1.2461 140 2.1267
1.593 1.4959 168 2.1148
1.604 1.7457 196 2.1129
1.6085 1.9955 224 2.1148

Framework versions

  • Transformers 4.45.1
  • Pytorch 2.4.1+cu121
  • Datasets 2.21.0
  • Tokenizers 0.20.0