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:
  - 4af3f7c1fa5610ea_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/4af3f7c1fa5610ea_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
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: lesso11/1f140cc0-8ab6-497b-b951-0e8bcc772491
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.000211
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/4af3f7c1fa5610ea_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: 110
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: 1798553a-3812-4302-9a96-9f20992e85cb
wandb_project: 11a
wandb_run: your_name
wandb_runid: 1798553a-3812-4302-9a96-9f20992e85cb
warmup_steps: 100
weight_decay: 0.0
xformers_attention: null

1f140cc0-8ab6-497b-b951-0e8bcc772491

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: 11.7416

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: 0.000211
• train_batch_size: 4
• eval_batch_size: 4
• seed: 110
• 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	11.9364
11.7938	0.1516	500	11.7862
11.7802	0.3033	1000	11.7740
11.7721	0.4549	1500	11.7675
11.771	0.6066	2000	11.7623
11.7686	0.7582	2500	11.7571
11.7635	0.9098	3000	11.7530
11.7783	1.0615	3500	11.7503
11.749	1.2131	4000	11.7483
11.7695	1.3648	4500	11.7463
11.7511	1.5164	5000	11.7456
11.7489	1.6681	5500	11.7442
11.7429	1.8197	6000	11.7433
11.7575	1.9713	6500	11.7425
11.7575	2.1230	7000	11.7422
11.7637	2.2746	7500	11.7419
11.7554	2.4263	8000	11.7417
11.7528	2.5779	8500	11.7416
11.7544	2.7295	9000	11.7416

Framework versions

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