See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/tinyllama-chat
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 3a1bcb8cca7edd27_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/3a1bcb8cca7edd27_train_data.json
  type:
    field_input: knowledge
    field_instruction: instruction
    field_output: response
    format: '{instruction} {input}'
    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: Romain-XV/879525cd-e4f2-4e44-8072-ec8f73b38ea4
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: 128
lora_dropout: 0.3
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 4140
micro_batch_size: 4
mlflow_experiment_name: /tmp/3a1bcb8cca7edd27_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
use_rslora: true
val_set_size: 0.01668335001668335
wandb_entity: null
wandb_mode: online
wandb_name: b82dc40a-1f45-42af-bd37-f4c50d3f06b9
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: b82dc40a-1f45-42af-bd37-f4c50d3f06b9
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

879525cd-e4f2-4e44-8072-ec8f73b38ea4

This model is a fine-tuned version of unsloth/tinyllama-chat on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6455

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: 4140

Training results

Training Loss	Epoch	Step	Validation Loss
0.9466	0.0001	1	0.8223
0.7105	0.0109	100	0.7313
0.681	0.0217	200	0.7192
0.7182	0.0326	300	0.7122
0.6349	0.0434	400	0.7070
0.742	0.0543	500	0.7019
0.6625	0.0652	600	0.6981
0.7204	0.0760	700	0.6951
0.7891	0.0869	800	0.6925
0.6709	0.0977	900	0.6898
0.6681	0.1086	1000	0.6867
0.6967	0.1194	1100	0.6846
0.7626	0.1303	1200	0.6823
0.7464	0.1412	1300	0.6798
0.6655	0.1520	1400	0.6773
0.6699	0.1629	1500	0.6752
0.808	0.1737	1600	0.6732
0.6473	0.1846	1700	0.6711
0.6322	0.1955	1800	0.6697
0.6771	0.2063	1900	0.6668
0.6453	0.2172	2000	0.6654
0.6398	0.2280	2100	0.6636
0.7477	0.2389	2200	0.6620
0.7543	0.2497	2300	0.6600
0.5852	0.2606	2400	0.6581
0.6464	0.2715	2500	0.6567
0.5976	0.2823	2600	0.6553
0.5494	0.2932	2700	0.6535
0.7006	0.3040	2800	0.6521
0.6583	0.3149	2900	0.6512
0.6454	0.3258	3000	0.6503
0.6695	0.3366	3100	0.6493
0.7171	0.3475	3200	0.6484
0.6111	0.3583	3300	0.6476
0.6028	0.3692	3400	0.6471
0.7963	0.3800	3500	0.6465
0.6989	0.3909	3600	0.6462
0.7354	0.4018	3700	0.6459
0.6742	0.4126	3800	0.6457
0.6839	0.4235	3900	0.6456
0.6608	0.4343	4000	0.6455
0.6948	0.4452	4100	0.6455

Framework versions

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