See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Qwen2.5-1.5B-Instruct
bf16: true
chat_template: llama3
cosine_min_lr_ratio: 0.3
dataset_prepared_path: null
datasets:
- data_files:
  - a88e78e41748bf83_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/a88e78e41748bf83_train_data.json
  type:
    field_instruction: prompt
    field_output: generation
    format: '{instruction}'
    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: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: Romain-XV/fb136667-94f4-4222-917a-d970c866edc3
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: 0.1
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
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 3947
micro_batch_size: 4
mlflow_experiment_name: /tmp/a88e78e41748bf83_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
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: 0.04
wandb_entity: null
wandb_mode: online
wandb_name: 895387c3-0be7-49d8-a314-abeba9f636b4
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 895387c3-0be7-49d8-a314-abeba9f636b4
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

fb136667-94f4-4222-917a-d970c866edc3

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: 0.7285

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: 4
• total_train_batch_size: 16
• 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: 3947

Training results

Training Loss	Epoch	Step	Validation Loss
1.8079	0.0005	1	1.7711
0.9176	0.0481	100	0.9042
0.8823	0.0963	200	0.8688
0.7735	0.1444	300	0.8444
0.8465	0.1926	400	0.8353
0.8342	0.2407	500	0.8236
0.8173	0.2888	600	0.8145
0.7584	0.3370	700	0.8094
0.8554	0.3851	800	0.8027
0.8138	0.4333	900	0.7981
0.7881	0.4814	1000	0.7917
0.842	0.5295	1100	0.7877
0.7291	0.5777	1200	0.7833
0.7927	0.6258	1300	0.7793
0.8393	0.6740	1400	0.7751
0.8374	0.7221	1500	0.7746
0.7801	0.7702	1600	0.7696
0.7769	0.8184	1700	0.7658
0.7168	0.8665	1800	0.7624
0.7109	0.9147	1900	0.7591
0.7169	0.9628	2000	0.7577
0.6319	1.0110	2100	0.7609
0.7444	1.0591	2200	0.7570
0.6964	1.1072	2300	0.7549
0.8156	1.1554	2400	0.7511
0.6715	1.2035	2500	0.7489
0.6032	1.2517	2600	0.7478
0.5837	1.2998	2700	0.7465
0.641	1.3479	2800	0.7438
0.7289	1.3961	2900	0.7430
0.7601	1.4442	3000	0.7404
0.7079	1.4924	3100	0.7381
0.7538	1.5405	3200	0.7365
0.7364	1.5886	3300	0.7352
0.7603	1.6368	3400	0.7343
0.5916	1.6849	3500	0.7315
0.6189	1.7331	3600	0.7309
0.5654	1.7812	3700	0.7292
0.7495	1.8293	3800	0.7290
0.6212	1.8775	3900	0.7285

Framework versions

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