See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: EleutherAI/pythia-160m
bf16: auto
chat_template: llama3
dataloader_num_workers: 6
dataset_prepared_path: null
datasets:
- data_files:
  - e3c7e63ded75b566_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/e3c7e63ded75b566_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:
  metric: eval_loss
  mode: min
  patience: 3
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
evals_per_epoch: null
flash_attention: true
fp16: true
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: false
group_by_length: true
hub_model_id: error577/1357afd1-d482-42a2-a0e1-74ca4cc84f8e
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0003
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.3
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 8
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 3000
micro_batch_size: 4
mlflow_experiment_name: /tmp/e3c7e63ded75b566_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 30
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: 20
sequence_len: 512
special_tokens:
  pad_token: <|endoftext|>
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.1
wandb_entity: null
wandb_mode: online
wandb_name: d29de543-689f-4e0c-ae46-c703463c14b2
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: d29de543-689f-4e0c-ae46-c703463c14b2
warmup_steps: 10
weight_decay: 0.01
xformers_attention: null

1357afd1-d482-42a2-a0e1-74ca4cc84f8e

This model is a fine-tuned version of EleutherAI/pythia-160m on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.7472

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: 0.0003
• 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: 3000

Training results

Training Loss	Epoch	Step	Validation Loss
24.5299	0.0036	1	3.3949
15.9629	0.3648	100	2.3421
17.5064	0.7296	200	2.3281
18.4084	1.0944	300	2.2702
17.3907	1.4592	400	2.1417
17.5815	1.8240	500	2.1559
22.7959	2.1888	600	2.3133
26.5806	2.5536	700	2.6053
19.992	2.9184	800	2.0848
15.7717	3.2832	900	2.0117
17.3197	3.6480	1000	2.0289
20.4746	4.0128	1100	2.0306
21.4667	4.3776	1200	2.1084
20.1942	4.7424	1300	1.9623
14.3553	5.1072	1400	1.8793
16.4309	5.4720	1500	2.0612
15.1784	5.8368	1600	1.9298
16.4193	6.2016	1700	1.9154
17.9363	6.5663	1800	1.8732
16.3783	6.9311	1900	1.8789
13.8409	7.2959	2000	1.8382
13.7778	7.6607	2100	1.8246
15.9805	8.0255	2200	1.7810
16.4949	8.3903	2300	1.7869
15.3153	8.7551	2400	1.7643
12.6055	9.1199	2500	1.7611
13.3154	9.4847	2600	1.7602
14.0602	9.8495	2700	1.7611
15.7382	10.2143	2800	1.7575
15.5844	10.5791	2900	1.7568
15.4755	10.9439	3000	1.7472

Framework versions

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