See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: peft-internal-testing/tiny-dummy-qwen2
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 2e65a200fe79612a_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/2e65a200fe79612a_train_data.json
  type:
    field_input: unmasked_text
    field_instruction: masked_text
    field_output: privacy_mask
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,3,4,5,6,7
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/ecbffcec-0f64-475a-ba6e-8fe210cff958
hub_repo: null
hub_strategy: null
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 53382
micro_batch_size: 4
mlflow_experiment_name: /tmp/2e65a200fe79612a_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
optimizer: adamw_bnb_8bit
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 100
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.024031413864203287
wandb_entity: null
wandb_mode: online
wandb_name: 362a31a3-5092-4387-8fb5-ca35f44c8b1a
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 362a31a3-5092-4387-8fb5-ca35f44c8b1a
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

ecbffcec-0f64-475a-ba6e-8fe210cff958

This model is a fine-tuned version of peft-internal-testing/tiny-dummy-qwen2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 11.8577

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: 0.0002
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• 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: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 53382

Training results

Training Loss	Epoch	Step	Validation Loss
11.9307	0.0002	1	11.9302
11.877	0.0158	100	11.8738
11.8702	0.0315	200	11.8661
11.8661	0.0473	300	11.8640
11.8668	0.0630	400	11.8624
11.8667	0.0788	500	11.8614
11.8646	0.0946	600	11.8608
11.8634	0.1103	700	11.8600
11.8634	0.1261	800	11.8596
11.8606	0.1418	900	11.8594
11.8596	0.1576	1000	11.8590
11.8657	0.1733	1100	11.8588
11.8643	0.1891	1200	11.8587
11.8607	0.2049	1300	11.8584
11.8615	0.2206	1400	11.8582
11.8622	0.2364	1500	11.8582
11.8617	0.2521	1600	11.8580
11.8592	0.2679	1700	11.8580
11.865	0.2837	1800	11.8580
11.8636	0.2994	1900	11.8577
11.8625	0.3152	2000	11.8576
11.8637	0.3309	2100	11.8576
11.8589	0.3467	2200	11.8577

Framework versions

• PEFT 0.13.2
• Transformers 4.46.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1