Spaces:
Running
Running
updates
Browse files
agent.py
CHANGED
@@ -16,10 +16,8 @@ prompt = """
|
|
16 |
[INSTRUCTIONS]
|
17 |
If the search results are irrelevant to the question respond with *** I do not have enough information to answer this question.***
|
18 |
Search results may include tables in a markdown format. When answering a question using a table be careful about which rows and columns contain the answer and include all relevant information from the relevant rows and columns that the query is asking about.
|
19 |
-
Do not cobble facts together from multiple search results, instead summarize the main facts into a consistent and easy to understand response.
|
20 |
Do not base your response on information or knowledge that is not in the search results.
|
21 |
Make sure your response is answering the query asked. If the query is related to an entity (such as a person or place), make sure you use search results related to that entity.
|
22 |
-
For queries where only a short answer is required, you can give a brief response.
|
23 |
Consider that each search result is a partial segment from a bigger text, and may be incomplete.
|
24 |
Your output should always be in a single language - the $vectaraLangName language. Check spelling and grammar for the $vectaraLangName language.
|
25 |
Search results for the query *** $vectaraQuery***, are listed below, some are text, some MAY be tables in markdown format.
|
@@ -56,14 +54,26 @@ def create_assistant_tools(cfg):
|
|
56 |
Responds to an user question about a particular result, based on the publications.
|
57 |
""",
|
58 |
tool_args_schema = QueryPublicationsArgs,
|
59 |
-
reranker = "multilingual_reranker_v1", rerank_k = 100,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
n_sentences_before = 2, n_sentences_after = 2, lambda_val = 0.005,
|
61 |
summary_num_results = 15,
|
62 |
vectara_summarizer = summarizer,
|
63 |
include_citations = True,
|
64 |
vectara_prompt_text=prompt,
|
65 |
save_history = True,
|
66 |
-
verbose=
|
67 |
)
|
68 |
|
69 |
search_publications = vec_factory.create_search_tool(
|
@@ -72,7 +82,19 @@ def create_assistant_tools(cfg):
|
|
72 |
Returns matching publications to a user query.
|
73 |
""",
|
74 |
tool_args_schema = QueryPublicationsArgs,
|
75 |
-
reranker = "
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
76 |
n_sentences_before = 2, n_sentences_after = 2, lambda_val = 0.005,
|
77 |
save_history = True,
|
78 |
verbose=True
|
|
|
16 |
[INSTRUCTIONS]
|
17 |
If the search results are irrelevant to the question respond with *** I do not have enough information to answer this question.***
|
18 |
Search results may include tables in a markdown format. When answering a question using a table be careful about which rows and columns contain the answer and include all relevant information from the relevant rows and columns that the query is asking about.
|
|
|
19 |
Do not base your response on information or knowledge that is not in the search results.
|
20 |
Make sure your response is answering the query asked. If the query is related to an entity (such as a person or place), make sure you use search results related to that entity.
|
|
|
21 |
Consider that each search result is a partial segment from a bigger text, and may be incomplete.
|
22 |
Your output should always be in a single language - the $vectaraLangName language. Check spelling and grammar for the $vectaraLangName language.
|
23 |
Search results for the query *** $vectaraQuery***, are listed below, some are text, some MAY be tables in markdown format.
|
|
|
54 |
Responds to an user question about a particular result, based on the publications.
|
55 |
""",
|
56 |
tool_args_schema = QueryPublicationsArgs,
|
57 |
+
# reranker = "multilingual_reranker_v1", rerank_k = 100,
|
58 |
+
reranker = "chain", rerank_k = 100,
|
59 |
+
rerank_chain = [
|
60 |
+
{
|
61 |
+
"type": "multilingual_reranker_v1",
|
62 |
+
# "cutoff": 0.2
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"type": "mmr",
|
66 |
+
"diversity_bias": 0.2,
|
67 |
+
"limit": 50
|
68 |
+
}
|
69 |
+
],
|
70 |
n_sentences_before = 2, n_sentences_after = 2, lambda_val = 0.005,
|
71 |
summary_num_results = 15,
|
72 |
vectara_summarizer = summarizer,
|
73 |
include_citations = True,
|
74 |
vectara_prompt_text=prompt,
|
75 |
save_history = True,
|
76 |
+
verbose=False
|
77 |
)
|
78 |
|
79 |
search_publications = vec_factory.create_search_tool(
|
|
|
82 |
Returns matching publications to a user query.
|
83 |
""",
|
84 |
tool_args_schema = QueryPublicationsArgs,
|
85 |
+
reranker = "chain", rerank_k = 100,
|
86 |
+
rerank_chain = [
|
87 |
+
{
|
88 |
+
"type": "multilingual_reranker_v1",
|
89 |
+
# "cutoff": 0.2
|
90 |
+
},
|
91 |
+
{
|
92 |
+
"type": "mmr",
|
93 |
+
"diversity_bias": 0.2,
|
94 |
+
"limit": 50
|
95 |
+
}
|
96 |
+
],
|
97 |
+
# reranker = "multilingual_reranker_v1", rerank_k = 100,
|
98 |
n_sentences_before = 2, n_sentences_after = 2, lambda_val = 0.005,
|
99 |
save_history = True,
|
100 |
verbose=True
|