See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: katuni4ka/tiny-random-olmo-hf
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - fd1aa1a857af2b5b_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/fd1aa1a857af2b5b_train_data.json
  type:
    field_input: system_msg
    field_instruction: prompt_msg
    field_output: truth
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
early_stopping_patience: 5
eval_max_new_tokens: 128
eval_steps: 200
eval_table_size: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 32
gradient_checkpointing: true
group_by_length: true
hub_model_id: error577/b6c7e253-5d8f-4e33-aed6-a9e781737b17
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: 128
lora_dropout: 0.15
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
micro_batch_size: 4
mlflow_experiment_name: /tmp/fd1aa1a857af2b5b_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
optimizer: adamw_torch_fused
output_dir: miner_id_24
pad_to_sequence_len: true
restore_best_weights: true
auto_resume_from_checkpoints: true
s2_attention: null
sample_packing: false
save_steps: 200
sequence_len: 1024
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.005
wandb_entity: null
wandb_mode: online
wandb_name: 17e50d10-bf6d-4181-9085-72485caf9329
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 17e50d10-bf6d-4181-9085-72485caf9329
warmup_steps: 10
weight_decay: 0.001
xformers_attention: null

b6c7e253-5d8f-4e33-aed6-a9e781737b17

This model is a fine-tuned version of katuni4ka/tiny-random-olmo-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 9.6182

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: 32
• total_train_batch_size: 128
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 10

Training results

Training Loss	Epoch	Step	Validation Loss
10.8193	0.0005	1	10.8338
9.7722	0.0944	200	9.7763
9.7568	0.1888	400	9.7539
9.7028	0.2832	600	9.7060
9.6697	0.3776	800	9.6646
9.6569	0.4719	1000	9.6443
9.6613	1.1327	1200	9.6330
9.5697	1.3215	1400	9.6291
9.6892	1.5102	1600	9.6253
9.5148	1.6990	1800	9.6233
9.6646	1.8878	2000	9.6218
9.5204	2.0766	2200	9.6212
9.636	2.2654	2400	9.6206
9.575	2.4541	2600	9.6204
9.6332	2.6429	2800	9.6203
9.6241	2.8317	3000	9.6202
9.6232	6.0491	3200	9.6204
9.6231	6.4266	3400	9.6197
9.6291	6.8042	3600	9.6193
9.6054	7.1818	3800	9.6189
9.5257	7.5593	4000	9.6186
9.559	7.9369	4200	9.6185
9.6064	8.3144	4400	9.6184
9.6427	8.6920	4600	9.6183
9.6201	9.0696	4800	9.6183
9.6109	9.4471	5000	9.6182
9.6314	9.8247	5200	9.6182

Framework versions

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