Raj Jayendrakumar Muchhala commited on
Commit
2a09216
·
1 Parent(s): 2ef1f15

change the output format and prompt for clipper

Browse files
Files changed (1) hide show
  1. clipper_prompts.py +15 -11
clipper_prompts.py CHANGED
@@ -7,25 +7,29 @@ Your task is to extract verbatim segments from the transcript for a given clip p
7
  - **Duration Target:** The desired duration for the clip.
8
 
9
  Instructions to follow when extracting verbatim segments from the transcript are as follows:
10
- 1. Return the transcript text exactly as it appears.
11
- 2. Identify and extract every segment of the transcript that relates to the Focus Prompt.
12
- 3. If there are multiple relevant passages, combine them in the order they appear in the transcript.
13
- 4. Duration Rules:
14
  - Use the Duration Target as a guideline to select the most relevant content.
15
  - Aim to extract transcript segments that roughly match the target duration.
16
- - If the available relevant content is naturally shorter than the duration target, extract only what is relevant, do not force inclusion of irrelevant content solely to reach the target duration.
 
17
 
18
  ### **OUTPUT:**
19
- - The response should be in the form of a JSON object with the following structure:
20
  ```
21
  {
22
- "Title": "Clip Title",
23
- "Focus Prompt": "Clip Focus Prompt",
24
- "Transcript": "Verbatim Segments",
25
- "Duration": "Clip Duration",
 
 
 
 
 
26
  }
27
  ```
28
-
29
  '''
30
 
31
  CLIPPER_USER_MESSAGE = '''
 
7
  - **Duration Target:** The desired duration for the clip.
8
 
9
  Instructions to follow when extracting verbatim segments from the transcript are as follows:
10
+ 1. Identify and extract every contiguous segment from the transcript that directly relates to the Focus Prompt.
11
+ 2. If there are multiple relevant passages, **do not** merge them into a single string. Instead, present them in sequential order as **separate JSON array elements**, each containing one contiguous passage from the transcript.
12
+ 3. Duration Rules:
 
13
  - Use the Duration Target as a guideline to select the most relevant content.
14
  - Aim to extract transcript segments that roughly match the target duration.
15
+ - If the available relevant content is naturally shorter than the duration target, extract only what is relevant, do not force inclusion of irrelevant content solely to reach the target duration.
16
+ 4. Return these passages **exactly** as they appear (verbatim) from the source.
17
 
18
  ### **OUTPUT:**
19
+ Return a JSON object with **only** the key `"Regions"`, where the value is an **array** of objects.
20
  ```
21
  {
22
+ "Regions": [
23
+ {
24
+ "region": "Exact verbatim text of the first relevant passage."
25
+ },
26
+ {
27
+ "region": "Exact verbatim text of the second relevant passage."
28
+ },
29
+ ....
30
+ ]
31
  }
32
  ```
 
33
  '''
34
 
35
  CLIPPER_USER_MESSAGE = '''