82 lines
2.8 KiB
Markdown
82 lines
2.8 KiB
Markdown
---
|
|
license: mit
|
|
datasets:
|
|
- GAIR/lima
|
|
language:
|
|
- en
|
|
pipeline_tag: text-generation
|
|
---
|
|
|
|
|
|
|
|
# lgaalves/gpt2-xl_lima (1.5B)
|
|
|
|
**lgaalves/lgaalves/gpt2-xl_lima** is an instruction fine-tuned model based on the GPT-2 transformer architecture.
|
|
|
|
|
|
### Benchmark Metrics
|
|
|
|
| Metric |gpt2-xl_lima |gpt2-xl (base) |
|
|
|-----------------------|-------|-------|
|
|
| Avg. | 36.65 | **36.66** |
|
|
| ARC (25-shot) | **31.14** | 30.29 |
|
|
| HellaSwag (10-shot) | 51.28 | **51.38** |
|
|
| MMLU (5-shot) | 25.43 | **26.43** |
|
|
| TruthfulQA (0-shot) | **38.74** | 38.54 |
|
|
|
|
|
|
We use state-of-the-art [Language Model Evaluation Harness](https://github.com/EleutherAI/lm-evaluation-harness) to run the benchmark tests above, using the same version as the HuggingFace LLM Leaderboard. Please see below for detailed instructions on reproducing benchmark results.
|
|
|
|
### Model Details
|
|
|
|
* **Trained by**: Luiz G A Alves
|
|
* **Model type:** **lgaalves/gpt2-xl_lima** is an auto-regressive language model based on the GPT-2 transformer architecture.
|
|
* **Language(s)**: English
|
|
|
|
### How to use:
|
|
|
|
```python
|
|
# Use a pipeline as a high-level helper
|
|
>>> from transformers import pipeline
|
|
>>> pipe = pipeline("text-generation", model="lgaalves/gpt2-xl_lima")
|
|
>>> question = "What is a large language model?"
|
|
>>> answer = pipe(question)
|
|
>>> print(answer[0]['generated_text'])
|
|
|
|
```
|
|
|
|
or, you can load the model direclty using:
|
|
|
|
```python
|
|
# Load model directly
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained("lgaalves/gpt2-xl_lima")
|
|
model = AutoModelForCausalLM.from_pretrained("lgaalves/gpt2-xl_lima")
|
|
```
|
|
|
|
### Training Dataset
|
|
|
|
`lgaalves/gpt2-xl_lima` trained on the [GAIR/lima](https://huggingface.co/datasets/GAIR/lima).
|
|
|
|
### Training Procedure
|
|
|
|
`lgaalves/gpt2-xl_lima` was instruction fine-tuned using LoRA on 1 Tesla V100-SXM2-16GB. It took about 10 minutes to train it.
|
|
|
|
# Intended uses, limitations & biases
|
|
|
|
You can use the raw model for text generation or fine-tune it to a downstream task. The model was not extensively tested and may produce false information. It contains a lot of unfiltered content from the internet, which is far from neutral.
|
|
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
|
|
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_lgaalves__gpt2-xl_lima)
|
|
|
|
| Metric | Value |
|
|
|-----------------------|---------------------------|
|
|
| Avg. | 29.95 |
|
|
| ARC (25-shot) | 31.14 |
|
|
| HellaSwag (10-shot) | 51.28 |
|
|
| MMLU (5-shot) | 25.43 |
|
|
| TruthfulQA (0-shot) | 38.74 |
|
|
| Winogrande (5-shot) | 57.22 |
|
|
| GSM8K (5-shot) | 0.91 |
|
|
| DROP (3-shot) | 4.89 |
|