README.md

metadata

base_model: bigcode/starencoder
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - accuracy
model-index:
  - name: stack-edu-classifier-c
    results: []

stack-edu-classifier-c

This model is a fine-tuned version of bigcode/starencoder on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4660
• Precision: 0.4766
• Recall: 0.3489
• F1 Macro: 0.3689
• Accuracy: 0.5471
• F1 Binary Minimum3: 0.7028

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: 0.0003
• train_batch_size: 64
• eval_batch_size: 256
• seed: 0
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 128
• total_eval_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1 Macro	Accuracy	F1 Binary Minimum3
No log	0	0	5.6344	0.0030	0.1667	0.0058	0.0178	0
0.5079	1.4493	1000	0.5017	0.4119	0.3155	0.3188	0.5307	0.6809
0.5001	2.8986	2000	0.4960	0.4423	0.3275	0.3419	0.5208	0.7030
0.4739	4.3478	3000	0.4846	0.4462	0.3255	0.3353	0.5374	0.6863
0.4739	5.7971	4000	0.4784	0.4522	0.3223	0.3328	0.5354	0.6934
0.4508	7.2464	5000	0.4890	0.4492	0.3423	0.3540	0.5409	0.6770
0.4798	8.6957	6000	0.4746	0.4663	0.3353	0.3520	0.5308	0.7028
0.4613	10.1449	7000	0.4775	0.4707	0.3254	0.3405	0.5385	0.6900
0.4668	11.5942	8000	0.4934	0.4711	0.3347	0.3526	0.5149	0.7036
0.4657	13.0435	9000	0.4690	0.4797	0.3330	0.3496	0.5390	0.7002
0.4561	14.4928	10000	0.4676	0.4807	0.3375	0.3561	0.5398	0.7040
0.4597	15.9420	11000	0.4668	0.4788	0.3336	0.3497	0.5413	0.7037
0.4524	17.3913	12000	0.4680	0.4759	0.3353	0.3541	0.5370	0.7038
0.4674	18.8406	13000	0.4660	0.4766	0.3489	0.3689	0.5471	0.7028

Framework versions

• Transformers 4.43.4
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1