See axolotl config

axolotl version: 0.4.1

adapter: lora
base_model: echarlaix/tiny-random-PhiForCausalLM
bf16: auto
chat_template: llama3
dataset_prepared_path: null
datasets:
- data_files:
  - 7462b07f6259b24d_train_data.json
  ds_type: json
  format: custom
  path: /workspace/input_data/7462b07f6259b24d_train_data.json
  type:
    field_instruction: startphrase
    field_output: gold-ending
    format: '{instruction}'
    no_input_format: '{instruction}'
    system_format: '{system}'
    system_prompt: ''
debug: null
deepspeed: null
do_eval: true
early_stopping_patience: 3
eval_batch_size: 4
eval_max_new_tokens: 128
eval_steps: 500
evals_per_epoch: null
flash_attention: true
fp16: false
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 8
gradient_checkpointing: true
group_by_length: true
hub_model_id: lesso14/6a18706f-1f89-4af9-a79b-aa75bcf38fa7
hub_repo: null
hub_strategy: checkpoint
hub_token: null
learning_rate: 0.000214
load_in_4bit: false
load_in_8bit: false
local_rank: null
logging_steps: 50
lora_alpha: 128
lora_dropout: 0.15
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 64
lora_target_linear: true
lr_scheduler: cosine
max_grad_norm: 1.0
max_steps: 50000
micro_batch_size: 4
mlflow_experiment_name: /tmp/7462b07f6259b24d_train_data.json
model_type: AutoModelForCausalLM
num_epochs: 10
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: 500
saves_per_epoch: null
seed: 140
sequence_len: 1024
special_tokens:
  pad_token: <|endoftext|>
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: 9611c628-3f80-4127-8fd5-47e5a88912ed
wandb_project: 14a
wandb_run: your_name
wandb_runid: 9611c628-3f80-4127-8fd5-47e5a88912ed
warmup_steps: 100
weight_decay: 0.0
xformers_attention: null

6a18706f-1f89-4af9-a79b-aa75bcf38fa7

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

• Loss: 6.7381

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: 0.000214
• train_batch_size: 4
• eval_batch_size: 4
• seed: 140
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 100
• training_steps: 27536

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0004	1	6.9353
6.8024	0.1816	500	6.7948
6.7909	0.3632	1000	6.7805
6.7839	0.5447	1500	6.7707
6.7782	0.7263	2000	6.7665
6.7752	0.9079	2500	6.7620
6.7731	1.0896	3000	6.7585
6.7665	1.2712	3500	6.7563
6.7647	1.4528	4000	6.7555
6.7693	1.6343	4500	6.7542
6.7655	1.8159	5000	6.7526
6.7691	1.9975	5500	6.7526
6.7587	2.1792	6000	6.7495
6.7598	2.3608	6500	6.7491
6.7643	2.5424	7000	6.7475
6.7603	2.7240	7500	6.7470
6.7626	2.9055	8000	6.7467
6.7655	3.0872	8500	6.7455
6.7542	3.2688	9000	6.7447
6.7588	3.4504	9500	6.7444
6.7615	3.6320	10000	6.7439
6.7594	3.8136	10500	6.7434
6.7554	3.9951	11000	6.7427
6.7579	4.1769	11500	6.7426
6.7579	4.3584	12000	6.7423
6.7618	4.5400	12500	6.7421
6.7577	4.7216	13000	6.7415
6.758	4.9032	13500	6.7417
6.7564	5.0849	14000	6.7413
6.7575	5.2665	14500	6.7409
6.7554	5.4480	15000	6.7403
6.7527	5.6296	15500	6.7403
6.7581	5.8112	16000	6.7402
6.7663	5.9928	16500	6.7398
6.765	6.1745	17000	6.7397
6.7492	6.3561	17500	6.7396
6.7621	6.5377	18000	6.7392
6.7587	6.7192	18500	6.7391
6.7595	6.9008	19000	6.7388
6.7544	7.0825	19500	6.7388
6.76	7.2641	20000	6.7387
6.7539	7.4457	20500	6.7386
6.7574	7.6273	21000	6.7385
6.7517	7.8088	21500	6.7384
6.7531	7.9904	22000	6.7383
6.756	8.1721	22500	6.7381
6.754	8.3537	23000	6.7381
6.7558	8.5353	23500	6.7381
6.7533	8.7169	24000	6.7380
6.7583	8.8985	24500	6.7384
6.7586	9.0802	25000	6.7384
6.7559	9.2617	25500	6.7381

Framework versions

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