Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: transformers
|
3 |
+
license: apache-2.0
|
4 |
+
base_model:
|
5 |
+
- meta-llama/Llama-2-13b-chat-hf
|
6 |
+
language:
|
7 |
+
- en
|
8 |
+
pipeline_tag: text-generation
|
9 |
+
---
|
10 |
+
|
11 |
+
# **Introduction**
|
12 |
+
This is a corning domain-specific model that has been trained to support multi-turn conversations in chemistry-related domains.
|
13 |
+
|
14 |
+
### **Loading the Model**
|
15 |
+
|
16 |
+
Use the following Python code to load the model:
|
17 |
+
|
18 |
+
```python
|
19 |
+
import torch
|
20 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
21 |
+
|
22 |
+
tokenizer = AutoTokenizer.from_pretrained("nayohan/corningQA-llama2-13b-chat")
|
23 |
+
model = AutoModelForCausalLM.from_pretrained(
|
24 |
+
"nayohan/corningQA-llama2-13b-chat",
|
25 |
+
device_map="auto",
|
26 |
+
torch_dtype=torch.float16,
|
27 |
+
)
|
28 |
+
```
|
29 |
+
|
30 |
+
### **Generating Text**
|
31 |
+
|
32 |
+
To generate text, use the following Python code:
|
33 |
+
|
34 |
+
```python
|
35 |
+
text = """
|
36 |
+
You will be shown dialogues between Speaker 1 and Speaker 2. Please read and understand given Dialogue Session,
|
37 |
+
then complete the task under the guidance of Task Introduction.\n\n
|
38 |
+
Context: {context}
|
39 |
+
Dialogue Session: {diaogues}
|
40 |
+
Task Introduction: After reading the Dialogue Session, please create an appropriate response in the parts marked ###.
|
41 |
+
Task Result:
|
42 |
+
"""
|
43 |
+
inputs = tokenizer(text, return_tensors="pt")
|
44 |
+
|
45 |
+
outputs = model.generate(**inputs, max_new_tokens=64)
|
46 |
+
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
47 |
+
```
|
48 |
+
|
49 |
+
```
|
50 |
+
# examples
|
51 |
+
{context}: A New CCn Inverter Circuit For AMLCD Panels Results In Significantly Higher Efficiency And Brightness Mehmet K. Nalbant Linfinity Microelectronics 11861 Western Ave., Garden Grove, CA 92641 Tel: (714) 898-8121 Fax: (714) 893-2570 Abstract - The paper will discuss a new CCFL inverter that is based on a new control technique named the Current Synchronous Zero Voltage Switching (CS-ZVS) topology. This topology enables the design of very eficient dimmable CCFL inverter
|
52 |
+
{diaogues}: Speaker 1: What is the name of the new control technique used in the CCFL inverter?\nSpeaker 2: The new control technique is called Current Synchronous Zero Voltage Switching (CS-ZVS) topology.\nSpeaker 1: Can you tell me more about the CS-ZVS topology?\nSpeaker 2: Sure! The CS-ZVS topology is a new control technique that allows for the design of highly efficient and dimmable CCFL inverters.\nSpeaker 1: That sounds impressive. How does it achieve higher efficiency and brightness?\nSpeaker 2: Well, the CS-ZVS topology synchronizes the switching of the inverter with the current flow, resulting in lower losses and improved performance.\nSpeaker 1: I see. So, is this new CCFL inverter already available in the market?\nSpeaker 2: Yes, it is. It was developed by Linfinity Microelectronics and is currently being used in AMLCD panels to improve their efficiency and brightness.\nSpeaker 1: That's amazing. I'm sure this new inverter will be very popular in the market.\nSpeaker 2: ###
|
53 |
+
```
|
54 |
+
|
55 |
+
### **License**
|
56 |
+
- [meta-llama/Llama-2-13b-chat-hf](https://llama.meta.com/llama-downloads/): Meta license
|
57 |
+
|
58 |
+
```bibtex
|
59 |
+
@article{touvron2023llama,
|
60 |
+
title={Llama 2: Open foundation and fine-tuned chat models},
|
61 |
+
author={Touvron, Hugo and Martin, Louis and Stone, Kevin and Albert, Peter and Almahairi, Amjad and Babaei, Yasmine and Bashlykov, Nikolay and Batra, Soumya and Bhargava, Prajjwal and Bhosale, Shruti and others},
|
62 |
+
journal={arXiv preprint arXiv:2307.09288},
|
63 |
+
year={2023}
|
64 |
+
}
|
65 |
+
```
|