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:
  - 41cfaa113d837568_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/41cfaa113d837568_train_data.json
  type:
    field_instruction: premise
    field_output: hypothesis
    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: false
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/c9fe4bf0-c690-4c16-939a-e900e96b8da2
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: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lora_target_modules:
- query_key_value
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 8832
micro_batch_size: 4
mlflow_experiment_name: /tmp/41cfaa113d837568_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: 0.029732703000029732
wandb_entity: null
wandb_mode: online
wandb_name: 791c5a53-d2c7-4e74-a1e0-6065f33c468d
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 791c5a53-d2c7-4e74-a1e0-6065f33c468d
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

c9fe4bf0-c690-4c16-939a-e900e96b8da2

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: 2.0059

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
24.65	0.0002	1	3.3734
17.8944	0.0196	100	2.3096
17.1278	0.0392	200	2.2631
20.0682	0.0588	300	2.2257
17.2783	0.0784	400	2.2106
20.953	0.0981	500	2.1982
18.0772	0.1177	600	2.1857
18.651	0.1373	700	2.1725
18.3775	0.1569	800	2.1652
18.2149	0.1765	900	2.1505
16.9199	0.1961	1000	2.1489
15.127	0.2157	1100	2.1384
18.3053	0.2353	1200	2.1365
16.8984	0.2550	1300	2.1276
16.4964	0.2746	1400	2.1240
15.2356	0.2942	1500	2.1224
17.4402	0.3138	1600	2.1186
16.7051	0.3334	1700	2.1171
15.4807	0.3530	1800	2.1089
16.4821	0.3726	1900	2.1048
17.9296	0.3922	2000	2.0956
20.2905	0.4118	2100	2.0972
18.5316	0.4315	2200	2.0932
17.8267	0.4511	2300	2.0877
16.5729	0.4707	2400	2.0820
18.0547	0.4903	2500	2.0832
16.7011	0.5099	2600	2.0749
16.4063	0.5295	2700	2.0728
15.8053	0.5491	2800	2.0709
18.0942	0.5687	2900	2.0662
16.7752	0.5884	3000	2.0613
16.2293	0.6080	3100	2.0576
18.3454	0.6276	3200	2.0525
14.8829	0.6472	3300	2.0511
15.7294	0.6668	3400	2.0501
17.0917	0.6864	3500	2.0480
17.7716	0.7060	3600	2.0442
15.2095	0.7256	3700	2.0400
17.79	0.7452	3800	2.0327
17.0274	0.7649	3900	2.0350
16.1242	0.7845	4000	2.0327
16.3873	0.8041	4100	2.0258
16.1772	0.8237	4200	2.0250
16.4274	0.8433	4300	2.0250
16.6272	0.8629	4400	2.0175
16.4568	0.8825	4500	2.0164
15.8856	0.9021	4600	2.0141
16.8868	0.9217	4700	2.0138
16.4577	0.9414	4800	2.0112
19.6963	0.9610	4900	2.0113
17.4742	0.9806	5000	2.0068
12.8209	1.0002	5100	2.0016
15.4832	1.0198	5200	2.0092
15.5734	1.0394	5300	2.0059

Framework versions

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