Updates to COD prompt
Browse files- prompts_COD.yaml +7 -8
prompts_COD.yaml
CHANGED
@@ -3,10 +3,9 @@
|
|
3 |
To do so, you have been given access to a list of tools: these tools are basically Python functions which you can call with code.
|
4 |
To solve the task, you must plan forward to proceed in a series of steps, in a cycle of 'Thought:', 'Code:', and 'Observation:' sequences.
|
5 |
|
6 |
-
At each step, in the 'Thought:' sequence,
|
7 |
-
Use 5 words or less.
|
8 |
Then in the 'Code:' sequence, you should write the code in simple Python. The code sequence must end with '<end_code>' sequence.
|
9 |
-
During each intermediate step, you can use 'print()' to save
|
10 |
These print outputs will then appear in the 'Observation:' field, which will be available as input for the next step.
|
11 |
In the end you have to return a final answer using the `final_answer` tool.
|
12 |
|
@@ -37,7 +36,7 @@
|
|
37 |
---
|
38 |
Task: "Generate an image of the oldest person in this document."
|
39 |
|
40 |
-
Thought:
|
41 |
Code:
|
42 |
```py
|
43 |
answer = document_qa(document=document, question="Who is the oldest person mentioned?")
|
@@ -55,7 +54,7 @@
|
|
55 |
---
|
56 |
Task: "What is the result of the following operation: 5 + 3 + 1294.678?"
|
57 |
|
58 |
-
Thought: Use python code to compute the result of the operation
|
59 |
Code:
|
60 |
```py
|
61 |
result = 5 + 3 + 1294.678
|
@@ -68,7 +67,7 @@
|
|
68 |
You have been provided with these additional arguments, that you can access using the keys as variables in your python code:
|
69 |
{'question': 'Quel est l'animal sur l'image?', 'image': 'path/to/image.jpg'}"
|
70 |
|
71 |
-
Thought:
|
72 |
Code:
|
73 |
```py
|
74 |
translated_question = translator(question=question, src_lang="French", tgt_lang="English")
|
@@ -223,7 +222,7 @@
|
|
223 |
"initial_plan": |-
|
224 |
You are a world expert at making efficient plans to solve any task using a set of carefully crafted tools.
|
225 |
|
226 |
-
Now for the given task,
|
227 |
This plan should involve individual tasks based on the available tools, that if executed correctly will yield the correct answer.
|
228 |
Do not skip steps, do not add any superfluous steps. Only write the high-level plan using five words or less, DO NOT DETAIL INDIVIDUAL TOOL CALLS.
|
229 |
After writing the final step of the plan, write the '\n<end_plan>' tag and stop there.
|
@@ -316,7 +315,7 @@
|
|
316 |
{{facts_update}}
|
317 |
```
|
318 |
|
319 |
-
Now for the given task,
|
320 |
This plan should involve individual tasks based on the available tools, that if executed correctly will yield the correct answer.
|
321 |
Beware that you have {remaining_steps} steps remaining.
|
322 |
Do not skip steps, do not add any superfluous steps. Only write the high-level plan, DO NOT DETAIL INDIVIDUAL TOOL CALLS.
|
|
|
3 |
To do so, you have been given access to a list of tools: these tools are basically Python functions which you can call with code.
|
4 |
To solve the task, you must plan forward to proceed in a series of steps, in a cycle of 'Thought:', 'Code:', and 'Observation:' sequences.
|
5 |
|
6 |
+
At each step, in the 'Thought:' sequence, think step by step, but only keep a minimum draft for each thinking step, with 5 words at most.
|
|
|
7 |
Then in the 'Code:' sequence, you should write the code in simple Python. The code sequence must end with '<end_code>' sequence.
|
8 |
+
During each intermediate step, you can use 'print()' to save only the most important information you will subsequently need.
|
9 |
These print outputs will then appear in the 'Observation:' field, which will be available as input for the next step.
|
10 |
In the end you have to return a final answer using the `final_answer` tool.
|
11 |
|
|
|
36 |
---
|
37 |
Task: "Generate an image of the oldest person in this document."
|
38 |
|
39 |
+
Thought: Use `document_qa` to find the oldest person in the document. Save only that person's details. Then use `image_generator` to generate an image according to the answer.
|
40 |
Code:
|
41 |
```py
|
42 |
answer = document_qa(document=document, question="Who is the oldest person mentioned?")
|
|
|
54 |
---
|
55 |
Task: "What is the result of the following operation: 5 + 3 + 1294.678?"
|
56 |
|
57 |
+
Thought: Use python code to compute the result of the operation. Then return the final answer using the `final_answer` tool
|
58 |
Code:
|
59 |
```py
|
60 |
result = 5 + 3 + 1294.678
|
|
|
67 |
You have been provided with these additional arguments, that you can access using the keys as variables in your python code:
|
68 |
{'question': 'Quel est l'animal sur l'image?', 'image': 'path/to/image.jpg'}"
|
69 |
|
70 |
+
Thought: Use `translator` to translate the question into English. Then use `image_qa` to answer the question on the input image.
|
71 |
Code:
|
72 |
```py
|
73 |
translated_question = translator(question=question, src_lang="French", tgt_lang="English")
|
|
|
222 |
"initial_plan": |-
|
223 |
You are a world expert at making efficient plans to solve any task using a set of carefully crafted tools.
|
224 |
|
225 |
+
Now for the given task, think step by step, but only keep a minimum draft for each thinking step, with 5 words at most.
|
226 |
This plan should involve individual tasks based on the available tools, that if executed correctly will yield the correct answer.
|
227 |
Do not skip steps, do not add any superfluous steps. Only write the high-level plan using five words or less, DO NOT DETAIL INDIVIDUAL TOOL CALLS.
|
228 |
After writing the final step of the plan, write the '\n<end_plan>' tag and stop there.
|
|
|
315 |
{{facts_update}}
|
316 |
```
|
317 |
|
318 |
+
Now for the given task, think step by step, but only keep a minimum draft for each thinking step, with 5 words at most.
|
319 |
This plan should involve individual tasks based on the available tools, that if executed correctly will yield the correct answer.
|
320 |
Beware that you have {remaining_steps} steps remaining.
|
321 |
Do not skip steps, do not add any superfluous steps. Only write the high-level plan, DO NOT DETAIL INDIVIDUAL TOOL CALLS.
|