AlexHung29629/dfjbvheriuer

See axolotl config

axolotl version: 0.16.0.dev0

base_model: mistralai/Ministral-3-3B-Instruct-2512-BF16

# Enable to use mistral-common tokenizer
tokenizer_use_mistral_common: false


# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name

plugins:
  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin

datasets:
  - path: AlexHung29629/test_data_123
    type:
      system_prompt: ""
      field_system: system
      field_instruction: input
      field_output: output
      format: "{instruction}"
      no_input_format: "{instruction}"
unfrozen_parameters:
  - ^model.language_model.norm.weight$
  - ^model.language_model.layers.2[0-5].[.a-z_]+$
#model.language_model.layers.25.post_attention_layernorm.weight
dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./outputs/out
dataloader_num_workers: 0
sequence_len: 32768
sample_packing: true

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 2
optimizer: adamw_bnb_8bit
lr_scheduler: constant
max_grad_norm: 1.0
learning_rate: 2e-5

bf16: auto
tf32: false

gradient_checkpointing: true
resume_from_checkpoint:
logging_steps: 1
#flash_attention: true
flex_attention: true
scaling_softmax: true

warmup_ratio: 0.0
evals_per_epoch: 1
saves_per_epoch: 1

# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

outputs/out

This model is a fine-tuned version of mistralai/Ministral-3-3B-Instruct-2512-BF16 on the AlexHung29629/test_data_123 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: 2e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• training_steps: 24

Training results

Framework versions

• Transformers 5.3.0
• Pytorch 2.9.1+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2