See axolotl config

axolotl version: 0.4.1

adapter: qlora
auto_resume_from_checkpoints: true
base_model: unsloth/tinyllama-chat
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
dataset_prepared_path: null
datasets:
- data_files:
  - 17acfb9d6efbdd55_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/17acfb9d6efbdd55_train_data.json
  type:
    field_instruction: question
    field_output: answer
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 3
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: false
group_by_length: true
hub_model_id: error577/b863e77b-b534-42a0-94c9-311dad4b8a87
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 10
lora_alpha: 128
lora_dropout: 0.05
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: null
micro_batch_size: 2
mlflow_experiment_name: /tmp/17acfb9d6efbdd55_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
optimizer: adamw_torch_4bit
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: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.005
wandb_entity: null
wandb_mode: online
wandb_name: abb078a6-11b7-4027-a142-41054e53ca1a
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: abb078a6-11b7-4027-a142-41054e53ca1a
warmup_steps: 30
weight_decay: 0.0
xformers_attention: null

b863e77b-b534-42a0-94c9-311dad4b8a87

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: 0.4750

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH_4BIT 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: 30
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0002	1	0.8539
0.6521	0.0164	100	0.7284
0.6338	0.0328	200	0.6849
0.5859	0.0492	300	0.6505
0.6338	0.0656	400	0.6351
0.5513	0.0820	500	0.6273
0.5518	0.0984	600	0.6050
0.5859	0.1149	700	0.5943
0.5074	0.1313	800	0.5759
0.4992	0.1477	900	0.5714
0.5219	0.1641	1000	0.5584
0.4703	0.1805	1100	0.5566
0.4787	0.1969	1200	0.5496
0.5744	0.2133	1300	0.5442
0.5262	0.2297	1400	0.5347
0.4898	0.2461	1500	0.5358
0.4646	0.2625	1600	0.5315
0.4983	0.2789	1700	0.5236
0.4164	0.2953	1800	0.5266
0.4321	0.3118	1900	0.5231
0.4791	0.3282	2000	0.5171
0.403	0.3446	2100	0.5146
0.4185	0.3610	2200	0.5143
0.5147	0.3774	2300	0.5039
0.3894	0.3938	2400	0.5091
0.4499	0.4102	2500	0.5056
0.4924	0.4266	2600	0.5015
0.5006	0.4430	2700	0.4962
0.3921	0.4594	2800	0.4935
0.5255	0.4758	2900	0.4969
0.4371	0.4922	3000	0.4953
0.4467	0.5087	3100	0.4926
0.4724	0.5251	3200	0.4886
0.4128	0.5415	3300	0.4827
0.4412	0.5579	3400	0.4836
0.4208	0.5743	3500	0.4800
0.4362	0.5907	3600	0.4805
0.4851	0.6071	3700	0.4744
0.4869	0.6235	3800	0.4735
0.4309	0.6399	3900	0.4715
0.4399	0.6563	4000	0.4815
0.4201	0.6727	4100	0.4749
0.419	0.6891	4200	0.4750

Framework versions

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