Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

New schema: Add chat schema #679

Merged
merged 4 commits into from
May 31, 2024
Merged

New schema: Add chat schema #679

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a4e4d54
feat: add chat schema
patrickamadeus May 22, 2024
399ee8a
Update __init__.py
sabilmakbar May 31, 2024
b20b861
Update chat.py
sabilmakbar May 31, 2024
80c3efc
Merge branch 'master' into feat/chat-schema
sabilmakbar May 31, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions seacrowd/utils/constants.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@
pairs_features_score,
pairs_multi_features,
qa_features,
chat_features,
image_features,
image_multi_features,
imqa_features,
Expand Down Expand Up @@ -105,6 +106,7 @@ class Tasks(Enum):
# Multi Text Generation
DIALOGUE_SYSTEM = "DS"
E2E_TASK_ORIENTED_DIALOGUE = "TOD"
MULTI_TURN_CONVERSATION = "MTC"

# Self Supervised & Unsupervised Text
PROMPTING = "PRT"
Expand Down Expand Up @@ -246,6 +248,7 @@ class Licenses(Enum):
Tasks.TOKEN_LEVEL_LANGUAGE_IDENTIFICATION: "SEQ_LABEL",
Tasks.COMMONSENSE_REASONING: "QA",
Tasks.QUESTION_ANSWERING: "QA",
Tasks.MULTI_TURN_CONVERSATION: "CHAT",
Tasks.CONCEPT_ALIGNMENT_CLASSIFICATION: "PAIRS",
Tasks.NEXT_SENTENCE_PREDICTION: "PAIRS",
Tasks.TEXT_RETRIEVAL: "PAIRS",
Expand Down Expand Up @@ -313,6 +316,7 @@ class Licenses(Enum):
"KB": kb_features,
"TREE": tree_features,
"QA": qa_features,
"CHAT": chat_features,
"T2T": text2text_features,
"TEXT": text_features(),
"TEXT_MULTI": text_multi_features(),
Expand Down
4 changes: 3 additions & 1 deletion seacrowd/utils/schemas/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
from .pairs import features_with_continuous_label as pairs_features_score
from .pairs_multilabel import features as pairs_multi_features
from .qa import features as qa_features
from .chat import features as chat_features
from .image import features as image_features
from .image import multi_features as image_multi_features
from .imqa import features as imqa_features
Expand All @@ -28,6 +29,7 @@
"pairs_features_score",
"pairs_multi_features",
"qa_features",
"chat_features",
"image_features",
"image_multi_features",
"imqa_features",
Expand All @@ -42,4 +44,4 @@
"text2text_features",
"video_features",
"tod_features",
]
]
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: add whitespace at EoF

24 changes: 24 additions & 0 deletions seacrowd/utils/schemas/chat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,24 @@
"""
Conversational Chat Schema
"""
import datasets

features = datasets.Features(
{
"id": datasets.Value("string"),
"input": datasets.Sequence({
"role": datasets.ClassLabel(names=["system", "user", "assistant"]),
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

the role should be implemented as string instead of ClassLabel due to 2 reasons:

  1. There might be some cases where the role can be multiple for a single-sequence dialogues
  2. The actual data (when the examples are generated) will result in an integer instead of string

"content": datasets.Value("string"),
}),
"output": datasets.Value("string"),

# the schema of 'meta' aren't specified either to allow some flexibility
"meta": {}

# notes on how to use this field of 'meta'
# you can choose two of options:
# 1. defining as empty dict if you don't think it's usable in `_generate_examples`, or
# 2. defining meta as dict of key with intended colname meta and its val with dataset.Features class
# in `_info` Dataloader method then populate it with the values in `_general_examples` Dataloader method
}
)
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: add whitespace at EoF