See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Llama-3.2-1B-Instruct
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - b46dd4e902cfc1e4_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/b46dd4e902cfc1e4_train_data.json
  type:
    field_input: OriginalAddress1
    field_instruction: PermitTypeDesc
    field_output: Description
    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/8a754eb6-61c7-4a8b-8a54-e51054953379
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: 3081
micro_batch_size: 4
mlflow_experiment_name: /tmp/b46dd4e902cfc1e4_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: 2048
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.04141507011571371
wandb_entity: null
wandb_mode: online
wandb_name: 9f56acc4-b827-4168-93d7-573d28b9c6d6
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 9f56acc4-b827-4168-93d7-573d28b9c6d6
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

8a754eb6-61c7-4a8b-8a54-e51054953379

This model is a fine-tuned version of unsloth/Llama-3.2-1B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.4427

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: 3081

Training results

Training Loss	Epoch	Step	Validation Loss
5.218	0.0003	1	5.2484
1.7686	0.0277	100	1.9123
1.8062	0.0553	200	1.8047
1.6583	0.0830	300	1.7470
1.9395	0.1106	400	1.7195
1.5875	0.1383	500	1.6879
1.6136	0.1659	600	1.6653
2.0235	0.1936	700	1.6588
1.6806	0.2212	800	1.6360
1.6857	0.2489	900	1.6220
1.5208	0.2765	1000	1.6051
1.4817	0.3042	1100	1.5931
1.6838	0.3318	1200	1.6109
1.4782	0.3595	1300	1.5708
1.465	0.3871	1400	1.5545
1.4195	0.4148	1500	1.5440
1.945	0.4424	1600	1.5307
1.5934	0.4701	1700	1.5188
1.4071	0.4977	1800	1.5075
1.4952	0.5254	1900	1.4991
1.8531	0.5530	2000	1.4884
1.517	0.5807	2100	1.4800
1.2886	0.6083	2200	1.4734
1.6789	0.6360	2300	1.4662
1.7322	0.6636	2400	1.4590
1.6309	0.6913	2500	1.4538
1.5905	0.7189	2600	1.4490
1.5501	0.7466	2700	1.4461
1.6598	0.7742	2800	1.4442
1.2199	0.8019	2900	1.4432
1.5627	0.8295	3000	1.4427

Framework versions

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