See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: HuggingFaceH4/tiny-random-LlamaForCausalLM
bf16: true
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - cb98dc95479f5003_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/cb98dc95479f5003_train_data.json
  type:
    field_input: examples
    field_instruction: func_desc
    field_output: answers
    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: 4
gradient_checkpointing: true
group_by_length: false
hub_model_id: Romain-XV/aed83255-c734-41e6-a1f1-58310fe0efed
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.00025
load_best_model_at_end: true
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 64
lora_dropout: 0.1
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lora_target_modules:
- q_proj
- k_proj
- v_proj
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 2451
micro_batch_size: 4
mlflow_experiment_name: /tmp/cb98dc95479f5003_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 3
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: 2048
special_tokens:
  pad_token: </s>
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: 8f653c4e-a627-4fd2-a4e3-e603285e2746
wandb_project: Gradients-On-Demand
wandb_run: your_name
wandb_runid: 8f653c4e-a627-4fd2-a4e3-e603285e2746
warmup_steps: 10
weight_decay: 0.0
xformers_attention: null

aed83255-c734-41e6-a1f1-58310fe0efed

This model is a fine-tuned version of HuggingFaceH4/tiny-random-LlamaForCausalLM on the None dataset. It achieves the following results on the evaluation set:

• Loss: 10.2273

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: 0.00025
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: 2451

Training results

Training Loss	Epoch	Step	Validation Loss
10.3771	0.0004	1	10.3787
10.2749	0.0425	100	10.2711
10.2473	0.0850	200	10.2448
10.256	0.1275	300	10.2385
10.2264	0.1699	400	10.2356
10.2249	0.2124	500	10.2334
10.2188	0.2549	600	10.2319
10.241	0.2974	700	10.2310
10.24	0.3399	800	10.2304
10.228	0.3824	900	10.2296
10.2459	0.4249	1000	10.2292
10.228	0.4673	1100	10.2289
10.2212	0.5098	1200	10.2284
10.238	0.5523	1300	10.2282
10.228	0.5948	1400	10.2280
10.229	0.6373	1500	10.2278
10.2222	0.6798	1600	10.2278
10.2453	0.7223	1700	10.2275
10.2329	0.7647	1800	10.2275
10.2318	0.8072	1900	10.2274
10.2093	0.8497	2000	10.2274
10.2278	0.8922	2100	10.2273
10.2535	0.9347	2200	10.2273
10.249	0.9772	2300	10.2273
11.0039	1.0198	2400	10.2273

Framework versions

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