add readme
Browse files
README.md
CHANGED
|
@@ -1,3 +1,65 @@
|
|
| 1 |
---
|
| 2 |
-
license:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license:
|
| 3 |
+
- apache-2.0
|
| 4 |
+
- cc-by-nc-4.0
|
| 5 |
+
datasets: pszemraj/fleece2instructions-codealpaca
|
| 6 |
+
tags:
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
- instruct
|
| 9 |
+
- instructions
|
| 10 |
+
- code
|
| 11 |
+
metrics:
|
| 12 |
+
- rouge
|
| 13 |
+
language:
|
| 14 |
+
- en
|
| 15 |
---
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
# bart-large-code-instructiongen
|
| 19 |
+
|
| 20 |
+
Use this text2text model to find out what LLM instructions might be able to generate an arbitary piece of code!
|
| 21 |
+
|
| 22 |
+
This model is a fine-tuned version of [facebook/bart-large](https://huggingface.co/facebook/bart-large) on the `pszemraj/fleece2instructions-codealpaca dataset.
|
| 23 |
+
It achieves the following results on the evaluation set:
|
| 24 |
+
- Loss: 0.9222
|
| 25 |
+
- Rouge1: 62.0692
|
| 26 |
+
- Rouge2: 36.1947
|
| 27 |
+
- Rougel: 57.5128
|
| 28 |
+
- Rougelsum: 58.6613
|
| 29 |
+
- Gen Len: 31.0060
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
## Intended uses & limitations
|
| 33 |
+
|
| 34 |
+
🚨 **note:** as the authors elected to release the [original dataset](https://github.com/sahil280114/codealpaca) under `cc-by-nc`, the license carries over to this model and **cannot be used for commercial activity**.
|
| 35 |
+
|
| 36 |
+
Intended use: Research on domain adaptation and/or other improvements to LLMs by extending instruction:text data pairs.
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 6e-05
|
| 48 |
+
- train_batch_size: 16
|
| 49 |
+
- eval_batch_size: 8
|
| 50 |
+
- seed: 42
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- gradient_accumulation_steps: 2
|
| 53 |
+
- total_train_batch_size: 32
|
| 54 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 55 |
+
- lr_scheduler_type: cosine
|
| 56 |
+
- lr_scheduler_warmup_ratio: 0.03
|
| 57 |
+
- num_epochs: 3.0
|
| 58 |
+
|
| 59 |
+
### Training results
|
| 60 |
+
|
| 61 |
+
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len |
|
| 62 |
+
|:-------------:|:-----:|:----:|:---------------:|:-------:|:-------:|:-------:|:---------:|:-------:|
|
| 63 |
+
| 1.0914 | 1.0 | 563 | 1.0303 | 60.288 | 34.1884 | 55.9293 | 57.0714 | 30.6267 |
|
| 64 |
+
| 0.8688 | 2.0 | 1126 | 0.9333 | 61.0409 | 34.9823 | 56.4887 | 57.6662 | 31.7255 |
|
| 65 |
+
| 0.6773 | 3.0 | 1689 | 0.9222 | 62.0692 | 36.1947 | 57.5128 | 58.6613 | 31.0060 |
|