See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: unsloth/Qwen2.5-Math-1.5B
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 29f15a2625e7ff70_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/29f15a2625e7ff70_train_data.json
  type:
    field_input: phonemes
    field_instruction: text_description
    field_output: text
    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/5f41d88f-6034-46f3-a2f6-0d2fac5aa67c
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: 2520
micro_batch_size: 4
mlflow_experiment_name: /tmp/29f15a2625e7ff70_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: 0.03253196265330687
wandb_entity: null
wandb_mode: online
wandb_name: 9e1f4c11-aa11-47f8-b2c5-a540a4bd0c15
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 9e1f4c11-aa11-47f8-b2c5-a540a4bd0c15
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

5f41d88f-6034-46f3-a2f6-0d2fac5aa67c

This model is a fine-tuned version of unsloth/Qwen2.5-Math-1.5B on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1811

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: 2520

Training results

Training Loss	Epoch	Step	Validation Loss
4.5759	0.0002	1	4.7131
0.4532	0.0215	100	0.5427
0.4828	0.0430	200	0.4158
0.2992	0.0646	300	0.3530
0.3098	0.0861	400	0.3217
0.355	0.1076	500	0.2956
0.248	0.1291	600	0.2758
0.3585	0.1506	700	0.2605
0.207	0.1722	800	0.2486
0.2147	0.1937	900	0.2392
0.2319	0.2152	1000	0.2316
0.1584	0.2367	1100	0.2230
0.1843	0.2582	1200	0.2172
0.2187	0.2798	1300	0.2106
0.2544	0.3013	1400	0.2050
0.1892	0.3228	1500	0.2019
0.1285	0.3443	1600	0.1957
0.2587	0.3658	1700	0.1924
0.1679	0.3874	1800	0.1888
0.2051	0.4089	1900	0.1864
0.1689	0.4304	2000	0.1842
0.2918	0.4519	2100	0.1830
0.1747	0.4734	2200	0.1820
0.2062	0.4950	2300	0.1814
0.2223	0.5165	2400	0.1812
0.1874	0.5380	2500	0.1811

Framework versions

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