See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: fxmarty/tiny-dummy-qwen2
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 09e55685d8a15ab8_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/09e55685d8a15ab8_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
deepspeed: null
early_stopping_patience: 2
eval_max_new_tokens: 128
eval_steps: 100
eval_table_size: null
flash_attention: true
fp16: null
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: false
hub_model_id: romainnn/98f08490-4ab5-48f6-b045-ed42ffbef8f0
hub_repo: null
hub_strategy: checkpoint
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: 32
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 16
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
lr_scheduler: cosine
max_steps: 1980
micro_batch_size: 4
mlflow_experiment_name: /tmp/09e55685d8a15ab8_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: 1613
wandb_entity: null
wandb_mode: online
wandb_name: fdf695ea-b676-42ab-ac8c-b9652dfdf1eb
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: fdf695ea-b676-42ab-ac8c-b9652dfdf1eb
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

98f08490-4ab5-48f6-b045-ed42ffbef8f0

This model is a fine-tuned version of fxmarty/tiny-dummy-qwen2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 11.9133

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: 1980

Training results

Training Loss	Epoch	Step	Validation Loss
11.9306	0.0008	1	11.9321
11.9209	0.0826	100	11.9209
11.9184	0.1653	200	11.9188
11.9177	0.2479	300	11.9163
11.9164	0.3305	400	11.9155
11.9169	0.4131	500	11.9151
11.9148	0.4958	600	11.9147
11.9151	0.5784	700	11.9144
11.9151	0.6610	800	11.9141
11.9137	0.7436	900	11.9138
11.9134	0.8263	1000	11.9137
11.9111	0.9089	1100	11.9136
11.9128	0.9915	1200	11.9135
11.5218	1.0742	1300	11.9134
11.6552	1.1568	1400	11.9134
11.885	1.2394	1500	11.9134
11.6617	1.3220	1600	11.9133
11.5499	1.4047	1700	11.9133
12.7226	1.4873	1800	11.9133
11.4889	1.5699	1900	11.9133

Framework versions

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