See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: Qwen/Qwen1.5-0.5B
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 0a95f702ed0ba111_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/0a95f702ed0ba111_train_data.json
  type:
    field_instruction: hotel_name
    field_output: review
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,4
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 33
eval_table_size: null
flash_attention: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/30f415ad-785f-4c50-bfc0-fd83679b2e67
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: 0.3
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: 4480.0
micro_batch_size: 4
mlflow_experiment_name: /tmp/0a95f702ed0ba111_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: 33
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.00970613699630002
wandb_entity: null
wandb_mode: online
wandb_name: c2e95a13-9e49-4df8-a617-dd6eaea1d861
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: c2e95a13-9e49-4df8-a617-dd6eaea1d861
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

30f415ad-785f-4c50-bfc0-fd83679b2e67

This model is a fine-tuned version of Qwen/Qwen1.5-0.5B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.1405

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
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• total_eval_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: 4480

Training results

Training Loss	Epoch	Step	Validation Loss
4.7914	0.0003	1	5.0087
3.476	0.0083	33	3.5184
3.5926	0.0166	66	3.4487
3.1628	0.0248	99	3.4158
3.4542	0.0331	132	3.3890
3.267	0.0414	165	3.3700
3.3527	0.0497	198	3.3570
3.3666	0.0580	231	3.3427
3.3278	0.0662	264	3.3361
3.3172	0.0745	297	3.3298
3.2072	0.0828	330	3.3193
3.3111	0.0911	363	3.3097
3.361	0.0994	396	3.3061
3.4016	0.1076	429	3.2981
3.4181	0.1159	462	3.2938
3.2431	0.1242	495	3.2890
3.3349	0.1325	528	3.2805
3.2384	0.1408	561	3.2766
3.3593	0.1490	594	3.2727
3.2874	0.1573	627	3.2688
3.3397	0.1656	660	3.2665
3.2402	0.1739	693	3.2600
3.4213	0.1822	726	3.2574
3.3238	0.1904	759	3.2527
3.2454	0.1987	792	3.2513
3.4028	0.2070	825	3.2510
3.2184	0.2153	858	3.2463
3.2074	0.2236	891	3.2459
3.1794	0.2318	924	3.2399
3.3006	0.2401	957	3.2406
3.2269	0.2484	990	3.2344
3.2229	0.2567	1023	3.2332
3.1653	0.2650	1056	3.2299
3.2942	0.2732	1089	3.2257
3.2257	0.2815	1122	3.2251
3.3287	0.2898	1155	3.2214
3.2034	0.2981	1188	3.2185
3.3216	0.3064	1221	3.2170
3.2412	0.3146	1254	3.2146
3.2114	0.3229	1287	3.2131
3.2999	0.3312	1320	3.2119
3.146	0.3395	1353	3.2074
3.12	0.3478	1386	3.2037
3.32	0.3560	1419	3.2034
3.2218	0.3643	1452	3.2036
3.1861	0.3726	1485	3.1985
3.214	0.3809	1518	3.1977
3.2622	0.3892	1551	3.1965
3.2881	0.3974	1584	3.1955
3.1741	0.4057	1617	3.1914
3.2859	0.4140	1650	3.1916
3.3133	0.4223	1683	3.1899
3.1447	0.4306	1716	3.1878
3.227	0.4388	1749	3.1845
3.1352	0.4471	1782	3.1818
3.1099	0.4554	1815	3.1816
3.1703	0.4637	1848	3.1782
3.276	0.4720	1881	3.1773
3.1768	0.4802	1914	3.1766
3.1777	0.4885	1947	3.1761
3.1467	0.4968	1980	3.1738
3.2644	0.5051	2013	3.1723
3.1731	0.5134	2046	3.1710
3.2887	0.5216	2079	3.1686
3.0899	0.5299	2112	3.1667
3.2466	0.5382	2145	3.1649
3.1688	0.5465	2178	3.1625
3.1678	0.5548	2211	3.1613
3.1074	0.5630	2244	3.1598
3.1225	0.5713	2277	3.1584
3.1169	0.5796	2310	3.1560
3.1652	0.5879	2343	3.1559
3.139	0.5962	2376	3.1554
3.2114	0.6044	2409	3.1537
3.2001	0.6127	2442	3.1498
3.2482	0.6210	2475	3.1491
3.117	0.6293	2508	3.1480
3.1729	0.6376	2541	3.1463
3.1911	0.6458	2574	3.1451
3.1628	0.6541	2607	3.1435
3.1665	0.6624	2640	3.1428
3.0256	0.6707	2673	3.1401
3.1604	0.6790	2706	3.1409
3.0809	0.6872	2739	3.1405

Framework versions

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