See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: llamafactory/tiny-random-Llama-3
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 006c0b96ba232e69_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/006c0b96ba232e69_train_data.json
  type:
    field_instruction: input
    field_output: output_answer
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: romainnn/263f118d-7409-47bb-a345-a9c30c9e0d2e
hub_repo: null
hub_strategy: checkpoint
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: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 4224
micro_batch_size: 4
mlflow_experiment_name: /tmp/006c0b96ba232e69_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 2
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: 2048
special_tokens:
  pad_token: <|eot_id|>
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.043117572998051086
wandb_entity: null
wandb_mode: online
wandb_name: 74e35e73-fca6-4927-a4ca-1b2f031982ea
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 74e35e73-fca6-4927-a4ca-1b2f031982ea
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

263f118d-7409-47bb-a345-a9c30c9e0d2e

This model is a fine-tuned version of llamafactory/tiny-random-Llama-3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 11.6429

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: 4224

Training results

Training Loss	Epoch	Step	Validation Loss
11.7609	0.0003	1	11.7689
11.732	0.0288	100	11.7277
11.7374	0.0577	200	11.7240
11.6868	0.0865	300	11.7068
11.7167	0.1154	400	11.6954
11.7079	0.1442	500	11.6860
11.6892	0.1730	600	11.6792
11.6885	0.2019	700	11.6725
11.7026	0.2307	800	11.6682
11.6537	0.2595	900	11.6638
11.6476	0.2884	1000	11.6601
11.6809	0.3172	1100	11.6582
11.7041	0.3461	1200	11.6558
11.6425	0.3749	1300	11.6542
11.6837	0.4037	1400	11.6517
11.6708	0.4326	1500	11.6509
11.695	0.4614	1600	11.6494
11.6767	0.4902	1700	11.6490
11.6534	0.5191	1800	11.6481
11.643	0.5479	1900	11.6464
11.7032	0.5768	2000	11.6464
11.6812	0.6056	2100	11.6457
11.6539	0.6344	2200	11.6452
11.6759	0.6633	2300	11.6451
11.6824	0.6921	2400	11.6444
11.635	0.7210	2500	11.6442
11.6331	0.7498	2600	11.6436
11.6758	0.7786	2700	11.6432
11.6911	0.8075	2800	11.6429
11.6362	0.8363	2900	11.6431
11.666	0.8651	3000	11.6429

Framework versions

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