See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Qwen2.5-1.5B-Instruct
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - f498211ddfc39ad0_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/f498211ddfc39ad0_train_data.json
  type:
    field_instruction: text
    field_output: text_description
    format: '{instruction}'
    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/c678b481-0ad6-4257-a56c-4d91becf4293
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: 4140
micro_batch_size: 4
mlflow_experiment_name: /tmp/f498211ddfc39ad0_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: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.03253196265330687
wandb_entity: null
wandb_mode: online
wandb_name: c45098d1-819e-4d15-988b-1e7aa32b705d
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: c45098d1-819e-4d15-988b-1e7aa32b705d
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

c678b481-0ad6-4257-a56c-4d91becf4293

This model is a fine-tuned version of unsloth/Qwen2.5-1.5B-Instruct on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4520

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
4.3494	0.0002	1	4.3327
0.5927	0.0215	100	0.6349
0.608	0.0430	200	0.5796
0.5354	0.0646	300	0.5645
0.6403	0.0861	400	0.5615
0.5192	0.1076	500	0.5449
0.5448	0.1291	600	0.5466
0.5212	0.1506	700	0.5367
0.5418	0.1722	800	0.5362
0.5331	0.1937	900	0.5267
0.5409	0.2152	1000	0.5253
0.4625	0.2367	1100	0.5164
0.5296	0.2582	1200	0.5211
0.4505	0.2798	1300	0.5087
0.4866	0.3013	1400	0.5103
0.5375	0.3228	1500	0.5059
0.4452	0.3443	1600	0.5032
0.5125	0.3658	1700	0.4982
0.4881	0.3874	1800	0.4961
0.4994	0.4089	1900	0.4927
0.5012	0.4304	2000	0.4886
0.4679	0.4519	2100	0.4882
0.5002	0.4734	2200	0.4848
0.4924	0.4950	2300	0.4808
0.5217	0.5165	2400	0.4782
0.4588	0.5380	2500	0.4752
0.4663	0.5595	2600	0.4734
0.4613	0.5811	2700	0.4707
0.4497	0.6026	2800	0.4701
0.449	0.6241	2900	0.4665
0.5207	0.6456	3000	0.4634
0.4806	0.6671	3100	0.4610
0.4238	0.6887	3200	0.4595
0.461	0.7102	3300	0.4572
0.507	0.7317	3400	0.4574
0.4169	0.7532	3500	0.4551
0.4393	0.7747	3600	0.4542
0.4432	0.7963	3700	0.4534
0.4521	0.8178	3800	0.4527
0.4717	0.8393	3900	0.4524
0.483	0.8608	4000	0.4520
0.463	0.8823	4100	0.4520

Framework versions

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