Hugging Face's logo

allura-org
/
GPT-J-6b-Disco-Elysium

Text Generation
Transformers
Safetensors
gptj
conversational
Model card Files
xet
Community
GPT-J-6b-Disco-Elysium / README.md
Fizzarolli's picture
Fizzarolli
Upload folder using huggingface_hub
1bd8702 verified
preview code
|
raw
history blame
3.09 kB
metadata
library_name: transformers
license: apache-2.0
base_model: EleutherAI/gpt-j-6b
tags:
  - generated_from_trainer
datasets:
  - ToastyPigeon/disco-chat
model-index:
  - name: ckpts
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.6.0

base_model: EleutherAI/gpt-j-6b

load_in_8bit: false
load_in_4bit: false

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

#unsloth_lora_mlp: true
#unsloth_lora_qkv: true
#unsloth_lora_o: true

strict: false

datasets:
  - path: ToastyPigeon/disco-chat
    type: completion
    field: text
    split: train

dataset_prepared_path: last_run_prepared
#val_set_size: 0.02
output_dir: ./ckpts

sequence_len: 2048
sample_packing: false
pad_to_sequence_len: true

#wandb_project: teleut-7b-rp
#wandb_entity:
#wandb_watch:
#wandb_name:
#wandb_log_model: checkpoint

# mlflow configuration if you're using it
#mlflow_tracking_uri: https://public-tracking.mlflow-e00zzfjq11ky6jcgtv.backbone-e00bgn6e63256prmhq.msp.eu-north1.nebius.cloud
#mlflow_experiment_name: gpt-j-6b-disco-chat
#mlflow_run_name: v1
#hf_mlflow_log_artifacts: true

gradient_accumulation_steps: 4
micro_batch_size: 4
num_epochs: 2
optimizer: paged_ademamix_8bit
lr_scheduler: cosine
learning_rate: 1e-5

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

gradient_checkpointing: unsloth
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
logging_steps: 1
xformers_attention:
flash_attention: true

#deepspeed: deepspeed_configs/zero3_bf16.json

warmup_steps: 25
#evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
weight_decay: 0.01

special_tokens:
  pad_token: "<|endoftext|>"

ckpts

This model is a fine-tuned version of EleutherAI/gpt-j-6b on the ToastyPigeon/disco-chat dataset.

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: 4
  • total_train_batch_size: 16
  • optimizer: Use paged_ademamix_8bit and the args are: No additional optimizer arguments
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 25
  • num_epochs: 2.0

Training results

Framework versions

  • Transformers 4.48.1
  • Pytorch 2.5.0+cu124
  • Datasets 3.2.0
  • Tokenizers 0.21.0