See axolotl config

axolotl version: 0.4.1

adapter: qlora
auto_resume_from_checkpoints: false
base_model: bigscience/bloomz-560m
bf16: auto
chat_template: llama3
dataloader_num_workers: 12
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: 4
eval_max_new_tokens: 128
eval_steps: 200
eval_table_size: null
evals_per_epoch: null
flash_attention: false
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 4
gradient_checkpointing: true
group_by_length: true
hub_model_id: error577/6b3ffec3-8960-4633-9dd8-9344f00c879f
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 10
lora_alpha: 128
lora_dropout: 0.05
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: null
micro_batch_size: 2
mlflow_experiment_name: /tmp/0fc9bc0ad0d49381_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: 200
sequence_len: 512
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.002
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: 0.01
xformers_attention: null

6b3ffec3-8960-4633-9dd8-9344f00c879f

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: 2.2038

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0000	1	2.9006
9.4598	0.0032	200	2.4428
9.0193	0.0063	400	2.4032
9.1876	0.0095	600	2.3742
9.0006	0.0126	800	2.3595
8.471	0.0158	1000	2.3460
8.8708	0.0189	1200	2.3285
8.9666	0.0221	1400	2.3163
8.9781	0.0252	1600	2.3118
8.5027	0.0284	1800	2.3064
8.4862	0.0315	2000	2.2939
8.5331	0.0347	2200	2.2867
8.6053	0.0378	2400	2.2798
8.3577	0.0410	2600	2.2777
8.7329	0.0441	2800	2.2695
8.2758	0.0473	3000	2.2647
8.6138	0.0505	3200	2.2641
8.846	0.0536	3400	2.2620
8.4359	0.0568	3600	2.2576
8.7765	0.0599	3800	2.2510
8.6455	0.0631	4000	2.2499
8.6459	0.0662	4200	2.2417
8.6934	0.0694	4400	2.2431
8.8205	0.0725	4600	2.2390
8.5673	0.0757	4800	2.2372
8.3182	0.0788	5000	2.2328
8.8069	0.0820	5200	2.2327
8.3561	0.0851	5400	2.2285
8.4872	0.0883	5600	2.2258
8.9193	0.0914	5800	2.2191
8.6298	0.0946	6000	2.2243
8.764	0.0978	6200	2.2229
8.2044	0.1009	6400	2.2133
8.6036	0.1041	6600	2.2200
8.3937	0.1072	6800	2.2183
8.6775	0.1104	7000	2.2122
8.8183	0.1135	7200	2.2136
8.9629	0.1167	7400	2.2077
8.3259	0.1198	7600	2.2085
8.5166	0.1230	7800	2.2127
8.0724	0.1261	8000	2.2054
8.4386	0.1293	8200	2.2090
8.5342	0.1324	8400	2.2019
7.7389	0.1356	8600	2.2047
8.7464	0.1387	8800	2.2020
8.6213	0.1419	9000	2.2024
8.5823	0.1451	9200	2.2038

Framework versions

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