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:
  - 5ec56af215ed5f07_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/5ec56af215ed5f07_train_data.json
  type:
    field_input: system
    field_instruction: prompt
    field_output: output
    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/554ddbdb-d049-46d3-8276-a503464009fc
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: 5220
micro_batch_size: 4
mlflow_experiment_name: /tmp/5ec56af215ed5f07_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
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.02552348671247282
wandb_entity: null
wandb_mode: online
wandb_name: eb779c22-cfa0-4af5-9286-869e115c7a74
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: eb779c22-cfa0-4af5-9286-869e115c7a74
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

554ddbdb-d049-46d3-8276-a503464009fc

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: 1.3328

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: 5220

Training results

Training Loss	Epoch	Step	Validation Loss
1.3513	0.0002	1	1.5537
1.3822	0.0168	100	1.4283
1.3041	0.0335	200	1.4172
1.3014	0.0503	300	1.4119
1.4278	0.0671	400	1.4072
1.4759	0.0838	500	1.4041
1.4432	0.1006	600	1.4006
1.29	0.1173	700	1.3981
1.3526	0.1341	800	1.3956
1.4521	0.1509	900	1.3932
1.3494	0.1676	1000	1.3912
1.4652	0.1844	1100	1.3885
1.469	0.2012	1200	1.3858
1.4669	0.2179	1300	1.3839
1.3072	0.2347	1400	1.3823
1.3941	0.2514	1500	1.3799
1.4834	0.2682	1600	1.3778
1.474	0.2850	1700	1.3751
1.4082	0.3017	1800	1.3733
1.3994	0.3185	1900	1.3709
1.4107	0.3353	2000	1.3698
1.3675	0.3520	2100	1.3672
1.3649	0.3688	2200	1.3646
1.2428	0.3855	2300	1.3629
1.3796	0.4023	2400	1.3609
1.4432	0.4191	2500	1.3594
1.4027	0.4358	2600	1.3575
1.242	0.4526	2700	1.3554
1.3509	0.4694	2800	1.3536
1.3217	0.4861	2900	1.3518
1.358	0.5029	3000	1.3501
1.352	0.5196	3100	1.3486
1.317	0.5364	3200	1.3470
1.3059	0.5532	3300	1.3454
1.3219	0.5699	3400	1.3437
1.498	0.5867	3500	1.3425
1.2141	0.6035	3600	1.3412
1.3785	0.6202	3700	1.3401
1.3292	0.6370	3800	1.3390
1.351	0.6537	3900	1.3380
1.2896	0.6705	4000	1.3370
1.3161	0.6873	4100	1.3362
1.406	0.7040	4200	1.3355
1.2631	0.7208	4300	1.3349
1.1527	0.7376	4400	1.3344
1.284	0.7543	4500	1.3339
1.4111	0.7711	4600	1.3335
1.1486	0.7878	4700	1.3332
1.2732	0.8046	4800	1.3331
1.3727	0.8214	4900	1.3329
1.3242	0.8381	5000	1.3328
1.2864	0.8549	5100	1.3328
1.3982	0.8717	5200	1.3328

Framework versions

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