|
# GPT2 Keyword Based Lecture Generator |
|
|
|
## Model description |
|
|
|
GPT2 fine-tuned on the TED Talks Dataset (published under the Creative Commons BY-NC-ND license). |
|
|
|
## Intended uses |
|
|
|
Used to generate spoken-word lectures. |
|
|
|
### How to use |
|
|
|
Input text: |
|
|
|
<BOS> title <|SEP|> Some keywords <|SEP|> |
|
|
|
Keyword Format: "Main Topic"."Subtopic1","Subtopic2","Subtopic3" |
|
|
|
Code Example: |
|
``` |
|
prompt = <BOS> + title + \\ |
|
<|SEP|> + keywords + <|SEP|> |
|
|
|
generated = torch.tensor(tokenizer.encode(prompt)).unsqueeze(0) |
|
|
|
model.eval(); |
|
``` |
|
|