diff --git a/tests/rest/model_test_co.rest b/tests/rest/model_test_co.rest
index 1012a56b..1c36a5e5 100644
--- a/tests/rest/model_test_co.rest
+++ b/tests/rest/model_test_co.rest
@@ -1,12 +1,15 @@
@local_endpoint = http://localhost:8000
@access_key = EMPTY
-### 1. Scenario: ambiguous location
+### 1. Scenario: Turns (Single). Parameters (Single, Required). | ambiguous location
+### Tool Used: get_weather
### Expected behavior(s): ask clarification about location
### Status: Approved
+### Trials: 10/10
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
Content-Type: application/json
+
{
"model": "Arch-Function",
"messages": [
@@ -25,9 +28,11 @@ Content-Type: application/json
}
-### 2. Scenario: ambiguous location
+### 2. Scenario: Turns (Single). Parameters (Single, Required) | ambiguous location
+### Tool Used: get_weather
### Expected behavior(s): model should clarify location as well, not just unit
### Status: Needs work
+### Trials: 3/10 model ask location, 3/10 say cant do the request 4/10 ask only about unit
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
Content-Type: application/json
@@ -49,9 +54,11 @@ Content-Type: application/json
"top_k": 10
}
-### 3. Scenario: undefine stock symbol
+### 3. Scenario: Turns (Single). Parameters (Single, Required) | undefine stock symbol
+### Tool Used: get_stock_price
### Expected behavior(s): clarification on the symbol
### Status: Approved
+### Trials: 9/10 ask clarification, 1/10 hallucinate choose AAPL
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
Content-Type: application/json
@@ -73,10 +80,12 @@ Content-Type: application/json
"top_k": 10
}
-### 4. Scenario: ambiguous stock
+### 4. Scenario: Turns (Single). Parameters (Single, Required) | ambiguous stock
+### Tool Used: get_stock_price
### Expected behavior(s): clarification on the symbol
-### Note: model doesn't ask clarification questions, sometime hallucinate
+### Note: model doesn't ask clarification questions, hallucinate
### Status: Needs work
+### Trials: 9/10 hallucinate, 1/10 ask clarification
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
Content-Type: application/json
@@ -98,9 +107,10 @@ Content-Type: application/json
"top_k": 10
}
-### 5. Scenario: ambiguous spotify parameter
+### 5. Scenario: Turns (Single). Parameters (Single, Required) | ambiguous spotify parameter
+### Tool Used: spotify_api
### Expected behavior(s): clarification on the music type
-### Note:
+### Trials: 8/10 ask clarficaition, 2/10 hallucinate
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -124,13 +134,16 @@ Content-Type: application/json
}
-### 6. Scenario: ambiguous location
+### 6. Scenario: Turns (Multi). Parameters (2, Required) | ambiguous location
+### Tool Used: spotify_api
### Expected behavior(s): clarification on the music type/ get the correct location
### Note:
+### Trials: 10/10 get correct tool
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
Content-Type: application/json
+
{
"model": "Arch-Function",
"messages": [
@@ -158,10 +171,11 @@ Content-Type: application/json
-### 7. Scenario: spotify | ambiguous artist
+### 7. Scenario: Turns (Single). Parameters (1, Required) | spotify | ambiguous artist
+### Tool Used: spotify_api
### Expected behavior(s): clarification on the artist
-### Note:
-### Status: Approved
+### Trials: 5/10 ask clarification, 5/10 hallucinate
+### Status: Needs work
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
Content-Type: application/json
@@ -183,9 +197,11 @@ Content-Type: application/json
"top_k": 10
}
-### 8. Scenario: spotify | ambiguous keywords
+### 8. Scenario: Turns (Multi). Parameters (1, Required) | spotify | ambiguous keywords
+### Tool Used: spotify_api
### Expected behavior(s): her as the keyword
### Note: miss the keyword her in the parameters
+### Trials: 8/10 wrong keyword
### Status: Needs work
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -217,9 +233,10 @@ Content-Type: application/json
"top_k": 10
}
-### 9. Scenario: product | ambiguous product
+### 9. Scenario: Turns (Single). Parameters (1, Required) | ambiguous product
+### Tool Used: product_recommendation
### Expected behavior(s): clarification question
-### Note:
+### Trials: 10/10
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -235,21 +252,18 @@ Content-Type: application/json
{
"role": "user",
"content": "Is my phone good?"
- },
- {
- "role": "assistant",
- "content": "{\"required_functions\": [\"query_database\"], \"clarification\": \"Could you please provide me with the phone model and its specifications?\"}",
- "tool_calls": []
- },
+ }
],
"temperature": 0.6,
"top_p": 1.0,
"top_k": 10
}
-### 10. Scenario: transfer money | ambiguous parameter
+### 10. Scenario: Turns (Multi). Parameters (3, Required). transfer money | ambiguous parameter
+### Tool Used: transfer_money
### Expected behavior(s): clarification question | track correct parameters
-### Note: sometimes it confirms the information again
+### Note: 1/30 it confirms the information again
+### Trials: 10/10
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -292,9 +306,11 @@ Content-Type: application/json
-### 10. Scenario: sale | ambiguous location
+### 11. Scenario: Turns (Multi). Parameters (2, Required). | ambiguous parameter
+### Tool Used: sales_opportunity
### Expected behavior(s): clarification question | track correct parameters
### Note: it doesn't understand the correction of location
+### Trials: 5/10 use US as location
### Status: Needs work
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -313,12 +329,11 @@ Content-Type: application/json
},
{
"role": "assistant",
- "content": "{\"required_functions\": [\"sales_opportunity\"], \"clarification\": \"In which industry are you looking for sales opportunities?\"}",
- "tool_calls": []
+ "content": "{\"required_functions\": [\"sales_opportunity\"], \"clarification\": \"In which industry are you looking for sales opportunities?\"}"
},
{
"role": "user",
- "content": "Private Banking in Sweden"
+ "content": "Private Banking in Europe"
}
],
"temperature": 0.6,
@@ -326,44 +341,12 @@ Content-Type: application/json
"top_k": 10
}
-### 10. Scenario: sale | ambiguous location
-### Expected behavior(s): clarification question | track correct parameters
-### Note: model follows user request and chooose random
-### Status: Not sure
-### Tested By: Co Tran
-POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
-Content-Type: application/json
-{
- "model": "Arch-Function",
- "messages": [
- {
- "role": "system",
- "content": "You are a helpful assistant designed to assist with the user query by making one or more function calls if needed.\n\nYou are provided with function signatures within XML tags:\n\n{\"id\": \"sales_opportunity\", \"type\": \"function\", \"function\": {\"name\": \"sales_opportunity\", \"description\": \"Retrieve potential sales opportunities based for a particular industry type in a region.\", \"parameters\": {\"type\": \"object\", \"properties\": {\"region\": {\"type\": \"str\", \"description\": \"Geographical region to identify sales opportunities.\"}, \"industry\": {\"type\": \"str\", \"description\": \"Industry type.\"}, \"max_results\": {\"type\": \"int\", \"description\": \"Maximum number of sales opportunities to retrieve.\", \"default\": 20}}, \"required\": [\"region\", \"industry\"]}}}\n\n\nYour task is to decide which functions are needed and collect missing parameters if necessary.\n\nBased on your analysis, provide your response in one of the following JSON formats:\n1. If no functions are needed:\n```\n{\"response\": \"Your response text here\"}\n```\n2. If functions are needed but some required parameters are missing:\n```\n{\"required_functions\": [\"func_name1\", \"func_name2\", ...], \"clarification\": \"Text asking for missing parameters\"}\n```\n3. If functions are needed and all required parameters are available:\n```\n{\"tool_calls\": [{\"name\": \"func_name1\", \"arguments\": {\"argument1\": \"value1\", \"argument2\": \"value2\"}},... (more tool calls as required)]}\n```"
- },
- {
- "role": "user",
- "content": "get me opportunity in US"
- },
- {
- "role": "assistant",
- "content": "{\"required_functions\": [\"sales_opportunity\"], \"clarification\": \"In which industry are you looking for sales opportunities?\"}",
- "tool_calls": []
- }
- ,
- {
- "role": "user",
- "content": "I don't really care"
- }
- ],
- "temperature": 0.6,
- "top_p": 1.0,
- "top_k": 10
-}
-
-### 11. Scenario: sale | ambiguous location
+### 12. Scenario: Turns (Multi). Parameters (2, Required). | ambiguous parameter
+### Tool Used: sales_opportunity
### Expected behavior(s): clarification question | track correct parameters
### Note: model get the correct tool and paramether
+### Trials: 10/10 use US as location
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -396,9 +379,11 @@ Content-Type: application/json
"top_k": 10
}
-### 12. Scenario: sale | ambiguous location
-### Expected behavior(s): clarification question | track correct parameters
+### 12. Scenario: Turns (Multi). Parameters (2, Required) | ambiguous intention
+### Tool Used: sales_opportunity
+### Expected behavior(s): cannot perform latest request
### Note: model response correctly because no matching tool provided
+### Trials: 10/10
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -431,9 +416,11 @@ Content-Type: application/json
"top_k": 10
}
-### 13. Scenario: sale | ambiguous request | multiple incomplete request
+### 13. Scenario: Turns (Multi). Parameters (2, Required) | ambiguous request | multiple incomplete request
+### Tool Used: product_recommendation,place_order
### Expected behavior(s): clarification question | track correct parameters
### Note:
+### Trials: 10/10
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -494,9 +481,11 @@ Content-Type: application/json
-### 14. Scenario: product | ambiguous request | multiple incomplete request
+### 14. Scenario: Turns (Multi). Parameters (2, Required) | ambiguous request | multiple incomplete request
+### Tool Used: product_recommendation,place_order
### Expected behavior(s): clarification question | track correct parameters
### Note: hallucinated user id but track the correct function
+### Trials: 10/10
### Status: Approved
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -539,9 +528,11 @@ Content-Type: application/json
-### 15. Scenario: product | ambiguous request | multiple incomplete request
+### 15. Scenario: Turns (Multi). Parameters (2, Required) | ambiguous request | multiple incomplete request
+### Tool Used: product_recommendation,place_order
### Expected behavior(s): clarification question | track correct parameters
-### Note: include 2 function calls with correct parameters (wrong id) but don't know the user intent to remove 1 function
+### Note:
+### Trials: 10/10 include 2 function calls with correct parameters (wrong id) but don't know the user intent to remove 1 function
### Status: Needs work
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
@@ -573,10 +564,12 @@ Content-Type: application/json
"top_k": 10
}
-### 16. Scenario: product | ambiguous request | multiple incomplete request | change parameter
+### 16. Scenario: Turns (Multi). Parameters (2, Required) | ambiguous request | multiple incomplete request | change parameter
+### Tool Used: product_recommendation,place_order
### Expected behavior(s): clarification question | track correct parameters
### Note:
### Status: Approved
+### Trials: 10/10 correct parameters
### Tested By: Co Tran
POST {{local_endpoint}}/v1/chat/completions HTTP/1.1
Content-Type: application/json