Datasets:
a6188466
/
dia-intent-sequencer-robot-arm-dataset

Dataset card Data Studio Files Community
system
stringclasses
27 values
in
stringlengths
4
415
out
stringlengths
1
336
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: initialize_components slot: components previous_questions_and_answers: [] current_question: Which components do you want to initialize? current_user_answer: abort
ABORT()
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: initialize_components slot: components previous_questions_and_answers: [] current_question: Which components do you want to initialize? current_user_answer: all of them
["table", "gripper", "camera", "shelf"]
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: initialize_components slot: components previous_questions_and_answers: [] current_question: Which components do you want to initialize? current_user_answer: cancel the request and retrieve 20 screws of 44 mm instead
ABORT_WITH_NEW_INTENTS([retrieve_screw(count=20, length=44)])
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: retrieve_screw slot: count previous_questions_and_answers: [] current_question: How many screws do you need? current_user_answer: actually give me 5 10mm screws
5, PROPAGATE_SLOT(length=10)
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: retrieve_screw slot: count previous_questions_and_answers: [] current_question: How many screws do you need? current_user_answer: actually give me screws of 20 mm
ASK("How many screws do you need?"), PROPAGATE_SLOT(length=20)
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: retrieve_screw slot: count previous_questions_and_answers: [] current_question: how many screws do you need? current_user_answer: just a few
F("a few")
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: retrieve_screw slot: length previous_questions_and_answers: [] current_question: what length do you need? current_user_answer: 12mm
12
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: retrieve_screw slot: length previous_questions_and_answers: [] current_question: what length do you need? current_user_answer: Do you have any 8mm screws?
QUERY_USER("Do you have any 8mm screws?")
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: retrieve_screw slot: length previous_questions_and_answers: [] current_question: what length do you need? current_user_answer: I don't know. What do you have in stock?
QUERY_USER("What length do you have in stock?")
You are a precise slot resolver. You resolve one slot at a time based on the current resolution context, but the user may change or override the task. Here are the available intents: - intent: retrieve_screw description: retrieve screws from an inventory parameters: - name: count type: int description: number of screws optional: False - name: length type: int description: length of the screws optional: False - intent: organize description: organize the screws that are on the table parameters: [] - intent: initialize_components description: initialize one or more of the following components: table, gripper, camera, shelf parameters: - name: components type: list[str] description: one or more of the components optional: False If the user's answer does not directly resolve to a value, return a QUERY_FILL(...), QUERY_USER(...) or a follow-up ASK(...). You have access to the following sources that can be queried to fill in missing information using QUERY_FILL: - inventory: description: > The inventory of screws, including their length and count. Useful to answer user queries about screw specifications or quantities, such as to resolve 'all' when the user asks for 'give me all screws'.
resolution_context: intent: retrieve_screw slot: length previous_questions_and_answers: [] current_question: what length do you need? current_user_answer: what length do you have?
QUERY_USER("what length do you have?")