Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
---
|
|
|
1 |
+
# README: tinyChat Instruction-Based LLM
|
2 |
+
|
3 |
+
Introducing tinyChat, the instruction-based Large Language Model (LLM) that’s less than 1% the size of GPT-3.5. tinyChat is an open-source model under the Apache 2.0 license and based on Google’s Flan-T5-Large, a 770m parameter model. Although not as performant as larger models, tinyChat can perform a variety of NLP tasks such as summarization, question answering, and sentiment analysis.
|
4 |
+
|
5 |
+
tinyChat is available on the HuggingFace model hub and the code repository is on GitHub. While tinyChat is open-sourced, we do not recommend using it in a production setting in its current state.
|
6 |
+
|
7 |
+
|
8 |
+
## Use Cases
|
9 |
+
|
10 |
+
- Chatbots
|
11 |
+
- Summarization
|
12 |
+
- Sentiment analysis
|
13 |
+
- Q&A systems
|
14 |
+
- Text completion
|
15 |
+
- Language modeling
|
16 |
+
- Mobile applications
|
17 |
+
- Complementing larger LLMs
|
18 |
+
|
19 |
+
## Future Directions
|
20 |
+
|
21 |
+
- Improving model accuracy
|
22 |
+
- Reducing biases and toxicity
|
23 |
+
- Developing new datasets
|
24 |
+
- Collaborating with the open-source community
|
25 |
+
- Applying tinyChat to new domains
|
26 |
+
|
27 |
+
## Acknowledgements
|
28 |
+
|
29 |
+
We express our gratitude to OpenAI, Hugging Face, Microsoft Research, and the creators of the Pile, Alpaca, and Databricks 15k datasets for their contributions to the landscape of open-source machine learning and the advancement of generative AI.
|
30 |
+
|
31 |
+
## Running the Code
|
32 |
+
|
33 |
+
```python
|
34 |
+
import transformers
|
35 |
+
from transformers import PeftModel
|
36 |
+
|
37 |
+
model_name = "google/flan-t5-large"
|
38 |
+
peft_model_id = "ckpts_databricks_large"
|
39 |
+
tokenizer = transformers.AutoTokenizer.from_pretrained(model_name)
|
40 |
+
base_model = transformers.AutoModelForSeq2SeqLM.from_pretrained(model_name)
|
41 |
+
peft_model = PeftModel.from_pretrained(base_model, peft_model_id)
|
42 |
+
|
43 |
+
inputs = tokenizer("""[INSERT INSTRUCTION HERE]""", return_tensors="pt")
|
44 |
+
outputs = peft_model.generate(**inputs, max_length=300, do_sample=True)
|
45 |
+
print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
|
46 |
+
|
47 |
+
|
48 |
+
|
49 |
---
|
50 |
license: apache-2.0
|
51 |
---
|