See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: katuni4ka/tiny-random-qwen1.5-moe
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 43f734d01e4861d3_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/43f734d01e4861d3_train_data.json
  type:
    field_input: testcase
    field_instruction: instruction
    field_output: output
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
do_eval: true
early_stopping_patience: 3
eval_batch_size: 4
eval_max_new_tokens: 128
eval_steps: 500
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: true
hub_model_id: lesso16/417ee334-17a2-4d4c-8f00-b7b4325cce09
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.000216
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 50
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
max_grad_norm: 1.0
max_steps: 9000
micro_batch_size: 4
mlflow_experiment_name: /tmp/43f734d01e4861d3_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
optimizer: adamw_torch_fused
output_dir: miner_id_24
pad_to_sequence_len: true
resume_from_checkpoint: null
s2_attention: null
sample_packing: false
save_steps: 500
saves_per_epoch: null
seed: 160
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.05
wandb_entity: null
wandb_mode: online
wandb_name: faf9ea99-82da-4871-8caa-b5447c2da5f9
wandb_project: 16a
wandb_run: your_name
wandb_runid: faf9ea99-82da-4871-8caa-b5447c2da5f9
warmup_steps: 100
weight_decay: 0.0
xformers_attention: null

417ee334-17a2-4d4c-8f00-b7b4325cce09

This model is a fine-tuned version of katuni4ka/tiny-random-qwen1.5-moe on the None dataset. It achieves the following results on the evaluation set:

• Loss: 11.5387

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: 0.000216
• train_batch_size: 4
• eval_batch_size: 4
• seed: 160
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• 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: 100
• training_steps: 9000

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0003	1	11.9301
11.6342	0.1437	500	11.6307
11.6202	0.2875	1000	11.6048
11.5994	0.4312	1500	11.5861
11.5912	0.5749	2000	11.5730
11.5788	0.7186	2500	11.5651
11.5702	0.8624	3000	11.5571
11.6381	1.0061	3500	11.5519
11.5565	1.1498	4000	11.5485
11.5557	1.2936	4500	11.5465
11.5574	1.4373	5000	11.5441
11.5434	1.5810	5500	11.5425
11.5525	1.7248	6000	11.5413
11.5419	1.8685	6500	11.5404
11.6024	2.0122	7000	11.5395
11.5531	2.1559	7500	11.5390
11.5662	2.2997	8000	11.5389
11.5531	2.4434	8500	11.5387
11.5524	2.5871	9000	11.5387

Framework versions

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