README.md

---
license: mit
datasets:
- DarwinAnim8or/greentext
language:
- en
tags:
- fun
- greentext
widget:
- text: ">be me"
  example_title: "be me"
- text: ">be zoo keeper"
co2_eq_emissions:
  emissions: 30
  source: "https://mlco2.github.io/impact/#compute"
  training_type: "fine-tuning"
  geographical_location: "Oregon, USA"
  hardware_used: "1x T4, Google Colab"
---

# GPT-Greentext-1.5b
A finetuned version of [GPT2-XL](https://huggingface.co/gpt2-xl) on the 'greentext' dataset.
A demo is available [here](https://huggingface.co/spaces/DarwinAnim8or/GPT-Greentext-Playground)
The demo playground is recommended over the inference box on the right. 

This is the largest release of the "GPT-Greentext" model series. The other models can be found here:
* [355m size model](https://huggingface.co/DarwinAnim8or/GPT-Greentext-355m)
* [125m size model](https://huggingface.co/DarwinAnim8or/GPT-Greentext-125m)

# Training Procedure
This was trained on the 'greentext' dataset, on Google Colab.
This model was trained for 1 epoch with learning rate 1e-2.
Notably this uses the "prompt" and "completion" style jsonl file, rather than the plain text file found in the greentext dataset.
This nets somewhat better, mostly more consistent results. 

# Biases & Limitations
This likely contains the same biases and limitations as the original GPT2 that it is based on, and additionally heavy biases from the greentext dataset.
It should be noted that offensive or not PG-output is definitely possible and likely will happen.

# Intended Use
This model is meant for fun, nothing else.

# Noteworthy differences between this model and the others
This model tends to like no_repeat_ngram_size values of 1 or 2; whereas the other models in this series tend to prefer 3. 

# Sample Use
```python
#Import model:
from happytransformer import HappyGeneration
happy_gen = HappyGeneration("GPT2", "DarwinAnim8or/GPT-Greentext-1.5b")

#Set generation settings:
from happytransformer import GENSettings
args_top_k = GENSettingsGENSettings(no_repeat_ngram_size=1, do_sample=True, top_k=80, temperature=0.8, max_length=150, early_stopping=False)

#Generate a response:
result = happy_gen.generate_text(""">be me
>""", args=args_top_k)

print(result)
print(result.text)
```