See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: katuni4ka/tiny-random-qwen1.5-moe
bf16: true
chat_template: llama3
dataloader_num_workers: 24
dataset_prepared_path: null
datasets:
- data_files:
  - a5ffd4a12886ce24_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/a5ffd4a12886ce24_train_data.json
  type:
    field_input: thinking
    field_instruction: prompt
    field_output: answer
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
device_map: auto
do_eval: true
early_stopping_patience: 3
eval_batch_size: 2
eval_max_new_tokens: 128
eval_steps: 150
eval_table_size: null
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: false
group_by_length: true
hub_model_id: abaddon182/2ce0ecfa-37d0-42e8-af3b-db72dfa8bfe8
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.0002
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 50
lora_alpha: 64
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 3000
micro_batch_size: 2
mlflow_experiment_name: /tmp/a5ffd4a12886ce24_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
optim_args:
  adam_beta1: 0.9
  adam_beta2: 0.999
  adam_epsilon: 1e-8
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: 150
saves_per_epoch: null
sequence_len: 512
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: f66709ae-7dde-4697-98bf-305cdca7fc8a
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: f66709ae-7dde-4697-98bf-305cdca7fc8a
warmup_steps: 50
weight_decay: 0.0
xformers_attention: null

2ce0ecfa-37d0-42e8-af3b-db72dfa8bfe8

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: 11.7931

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: 8
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=adam_beta1=0.9,adam_beta2=0.999,adam_epsilon=1e-8
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 50
• training_steps: 3000

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0011	1	11.9408
11.8524	0.1613	150	11.8495
11.8421	0.3227	300	11.8400
11.8306	0.4840	450	11.8285
11.8243	0.6453	600	11.8227
11.8179	0.8067	750	11.8170
11.8154	0.9680	900	11.8122
11.7959	1.1296	1050	11.8081
11.8065	1.2909	1200	11.8047
11.7935	1.4523	1350	11.8017
11.8127	1.6136	1500	11.7996
11.7972	1.7749	1650	11.7975
11.7957	1.9363	1800	11.7967
11.7989	2.0979	1950	11.7957
11.7967	2.2592	2100	11.7954
11.7953	2.4205	2250	11.7942
11.7964	2.5819	2400	11.7939
11.7939	2.7432	2550	11.7934
11.7942	2.9045	2700	11.7935
11.7973	3.0661	2850	11.7933
11.8017	3.2275	3000	11.7931

Framework versions

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