See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: katuni4ka/tiny-random-qwen1.5-moe
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 781f565e52c3a483_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/781f565e52c3a483_train_data.json
  type:
    field_input: documents
    field_instruction: question
    field_output: answer
    format: '{instruction} {input}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
device_map:
  ? ''
  : 0,1,2,3,4,5,6,7
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: true
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: Alphatao/48d9fe36-e614-454c-a088-feef9b4bef26
hub_repo: null
hub_strategy: null
hub_token: null
learning_rate: 0.0002
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 128
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 11515
micro_batch_size: 4
mlflow_experiment_name: /tmp/781f565e52c3a483_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 2
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: 100
sequence_len: 1024
strict: false
tf32: true
tokenizer_type: AutoTokenizer
train_on_inputs: false
trust_remote_code: true
val_set_size: 0.04
wandb_entity: null
wandb_mode: online
wandb_name: e51e0b4d-614b-41d3-94d3-22faff54d2e5
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: e51e0b4d-614b-41d3-94d3-22faff54d2e5
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

48d9fe36-e614-454c-a088-feef9b4bef26

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: 11.8163

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: 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: 2385

Training results

Training Loss	Epoch	Step	Validation Loss
11.9275	0.0008	1	11.9320
11.8559	0.0839	100	11.8529
11.8449	0.1678	200	11.8425
11.8434	0.2516	300	11.8380
11.8311	0.3355	400	11.8331
11.8355	0.4194	500	11.8294
11.8316	0.5033	600	11.8267
11.8228	0.5871	700	11.8243
11.8213	0.6710	800	11.8227
11.8192	0.7549	900	11.8216
11.8255	0.8388	1000	11.8207
11.8323	0.9226	1100	11.8197
13.3461	1.0065	1200	11.8190
12.1266	1.0904	1300	11.8184
11.0089	1.1743	1400	11.8179
12.0953	1.2581	1500	11.8174
11.2672	1.3420	1600	11.8171
11.1938	1.4259	1700	11.8169
12.7096	1.5098	1800	11.8167
12.2526	1.5936	1900	11.8165
11.0318	1.6775	2000	11.8164
11.9244	1.7614	2100	11.8164
11.4206	1.8453	2200	11.8163
11.6137	1.9291	2300	11.8163

Framework versions

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