See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: bigscience/bloomz-560m
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 0fc9bc0ad0d49381_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/0fc9bc0ad0d49381_train_data.json
  type:
    field_input: input
    field_instruction: instruction
    field_output: output
    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: false
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Romain-XV/d156f76d-6606-4266-96da-a418e1a226c2
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: 8832
micro_batch_size: 4
mlflow_experiment_name: /tmp/0fc9bc0ad0d49381_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
use_rslora: true
val_set_size: 0.00983438889107431
wandb_entity: null
wandb_mode: online
wandb_name: 7af274cb-a9a3-45b8-b6c3-b1c837d298d4
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 7af274cb-a9a3-45b8-b6c3-b1c837d298d4
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

d156f76d-6606-4266-96da-a418e1a226c2

This model is a fine-tuned version of bigscience/bloomz-560m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0438

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: 8832

Training results

Training Loss	Epoch	Step	Validation Loss
23.4576	0.0001	1	2.8683
19.7937	0.0064	100	2.4114
20.4259	0.0127	200	2.3545
18.2863	0.0191	300	2.3214
19.4042	0.0254	400	2.2988
18.4537	0.0318	500	2.2798
15.74	0.0381	600	2.2650
18.8956	0.0445	700	2.2526
17.846	0.0509	800	2.2435
17.2099	0.0572	900	2.2312
18.1726	0.0636	1000	2.2239
17.4762	0.0699	1100	2.2156
17.8853	0.0763	1200	2.2092
18.1147	0.0826	1300	2.2030
17.74	0.0890	1400	2.1963
18.1605	0.0953	1500	2.1899
18.1325	0.1017	1600	2.1850
18.0614	0.1081	1700	2.1802
17.757	0.1144	1800	2.1778
17.6561	0.1208	1900	2.1716
18.2847	0.1271	2000	2.1688
17.6908	0.1335	2100	2.1631
18.241	0.1398	2200	2.1593
16.251	0.1462	2300	2.1577
17.7695	0.1526	2400	2.1521
17.2459	0.1589	2500	2.1489
18.3262	0.1653	2600	2.1450
16.7641	0.1716	2700	2.1425
16.4128	0.1780	2800	2.1371
15.7316	0.1843	2900	2.1358
17.9185	0.1907	3000	2.1317
16.555	0.1971	3100	2.1280
15.8804	0.2034	3200	2.1261
17.6585	0.2098	3300	2.1229
17.4634	0.2161	3400	2.1184
17.5052	0.2225	3500	2.1192
17.4755	0.2288	3600	2.1174
18.0033	0.2352	3700	2.1110
16.3309	0.2415	3800	2.1089
16.633	0.2479	3900	2.1069
17.9653	0.2543	4000	2.1034
16.6872	0.2606	4100	2.1017
16.6698	0.2670	4200	2.0987
17.0016	0.2733	4300	2.0968
17.7949	0.2797	4400	2.0948
16.2796	0.2860	4500	2.0921
17.2325	0.2924	4600	2.0895
17.4596	0.2988	4700	2.0868
17.2106	0.3051	4800	2.0839
17.0064	0.3115	4900	2.0823
15.9642	0.3178	5000	2.0800
17.6006	0.3242	5100	2.0779
17.3074	0.3305	5200	2.0746
16.0723	0.3369	5300	2.0735
16.5184	0.3433	5400	2.0711
16.2517	0.3496	5500	2.0701
17.1206	0.3560	5600	2.0683
17.2825	0.3623	5700	2.0668
16.9153	0.3687	5800	2.0644
16.2446	0.3750	5900	2.0628
15.8944	0.3814	6000	2.0610
17.7732	0.3877	6100	2.0603
17.8103	0.3941	6200	2.0587
15.7341	0.4005	6300	2.0580
15.6502	0.4068	6400	2.0557
16.8526	0.4132	6500	2.0548
17.1581	0.4195	6600	2.0530
16.0818	0.4259	6700	2.0520
15.5948	0.4322	6800	2.0514
16.6084	0.4386	6900	2.0505
16.8273	0.4450	7000	2.0496
15.6169	0.4513	7100	2.0491
18.0275	0.4577	7200	2.0479
17.1104	0.4640	7300	2.0470
17.2611	0.4704	7400	2.0465
15.66	0.4767	7500	2.0461
15.8305	0.4831	7600	2.0450
15.9643	0.4895	7700	2.0455
17.4456	0.4958	7800	2.0441
16.9549	0.5022	7900	2.0445
15.9483	0.5085	8000	2.0437
15.8849	0.5149	8100	2.0439
16.6432	0.5212	8200	2.0436
16.6686	0.5276	8300	2.0434
16.7146	0.5339	8400	2.0431
16.051	0.5403	8500	2.0432
15.9274	0.5467	8600	2.0438

Framework versions

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