before_code stringlengths 30 149k | reviewer_comment stringlengths 16 20.6k | after_code stringlengths 30 163k | diff_context stringlengths 0 22.4k | file_path stringlengths 6 155 | comment_line int64 0 26 | language stringclasses 19
values | quality_score float64 0.07 1 | comment_type stringclasses 9
values | comment_length int64 16 20.6k | before_lines int64 2 4.61k | after_lines int64 2 4.4k | is_negative bool 2
classes | pr_title stringlengths 7 140 | pr_number int64 18 61.2k | repo_name stringclasses 29
values | repo_stars int64 321 56k | repo_language stringclasses 14
values | reviewer_username stringclasses 378
values | author_username stringclasses 739
values | user stringlengths 361 200k | assistant stringlengths 16 20.6k | system stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
return (
<FormControl ref={containerRef}>
<Menu placement="auto">
<MenuButton as={Button} pr="3" rightIcon={<Cog />} size="lg" width="100%">
{activePluginIndex > -1 && plugins[activePluginIndex]?.enabled ? (
<Box display="flex" justifyContent="flex-start" gap={2}>
... | Can you revert this? Already fixed in my PR | return (
<FormControl ref={containerRef}>
<Menu placement="auto">
<MenuButton as={Button} pr="3" rightIcon={<Cog />} size="lg" width="100%">
{activePluginIndex > -1 && plugins[activePluginIndex]?.enabled ? (
<Box display="flex" justifyContent="flex-start" gap={2}>
... | @@ -130,8 +130,8 @@ export const PluginsChooser = ({ plugins }: { plugins: PluginEntry[] }) => {
</Box>
)}
</MenuButton>
- <Box position="fixed" zIndex="1">
- <MenuList left="-150px" position="absolute" w="280px">
+ <Box position="fixed" zIndex="2">
+ <Me... | website/src/components/Chat/PluginsChooser.tsx | 26 | TypeScript | 0.071 | suggestion | 43 | 51 | 51 | false | Plugins fixes | 3,070 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | draganjovanovich | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/PluginsChooser.tsx
PR Title: Plugins fixes
BEFORE CODE: return (
<FormControl ref={containerRef}>
<Menu placement="auto">
<MenuButton as={Button} pr="3" rightIcon={<Cog />} size="lg" width="100%">
{... | Can you revert this? Already fixed in my PR | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
"--anonymizer-seed",
type=int,
help="Seed for the anonymizer. If not specified, no anonymization will be performed.",
)
parser.add_argument(
"--from-date",
type=parse_date,
help="Only export chats created on or after this date. Format: YYYY-MM-DD",
)
parse... | We could just call this `--scored` for simplicity | help="Seed for the anonymizer. If not specified, no anonymization will be performed.",
)
parser.add_argument(
"--from-date",
type=parse_date,
help="Only export chats created on or after this date. Format: YYYY-MM-DD",
)
parser.add_argument(
"--to-date",
ty... | @@ -219,19 +228,68 @@ def parse_args() -> argparse.Namespace:
type=int,
help="Seed for the anonymizer. If not specified, no anonymization will be performed.",
)
- # TODO: filters: reported, score, user ID, chat ID, etc
- # TODO: date range?
+ parser.add_argument(
+ "--from-date",
... | inference/server/export.py | 26 | Python | 0.286 | suggestion | 49 | 51 | 51 | false | Improve inference export script | 3,071 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | simonusher | Review this pull request change.
Language: Python
File: inference/server/export.py
PR Title: Improve inference export script
BEFORE CODE: "--anonymizer-seed",
type=int,
help="Seed for the anonymizer. If not specified, no anonymization will be performed.",
)
parser.add_argument(
... | We could just call this `--scored` for simplicity | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
if not file:
raise RuntimeError("File name must be specified when using compression.")
out_buff = gzip.open(file, "wt", encoding="UTF-8")
else:
out_buff = smart_open(file)
with out_buff as f:
for chat in chats:
if write_trees:
export_chat ... | This isn't really to do with your PR, but could you also add `not_(DbChat.deleted)` here? I forgot to include it when originally developing this script | if not file:
raise RuntimeError("File name must be specified when using compression.")
out_buff = gzip.open(file, "wt", encoding="UTF-8")
else:
out_buff = smart_open(file)
with out_buff as f:
for chat in chats:
if write_trees:
export_chat ... | @@ -163,13 +164,16 @@ def write_messages_to_file(
f.write("\n")
-async def fetch_eligible_chats(session_generator) -> list[DbChat]:
- """Fetch chats which are not opted out of data collection."""
+async def fetch_eligible_chats(
+ session_generator, filters: list[sqlalchemy.sql.elements.B... | inference/server/export.py | 26 | Python | 0.571 | suggestion | 151 | 51 | 51 | false | Improve inference export script | 3,071 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | simonusher | Review this pull request change.
Language: Python
File: inference/server/export.py
PR Title: Improve inference export script
BEFORE CODE: if not file:
raise RuntimeError("File name must be specified when using compression.")
out_buff = gzip.open(file, "wt", encoding="UTF-8")
else:
... | This isn't really to do with your PR, but could you also add `not_(DbChat.deleted)` here? I forgot to include it when originally developing this script | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
type=parse_date,
help="Only export chats created on or after this date. Format: YYYY-MM-DD",
)
parser.add_argument(
"--to-date",
type=parse_date,
help="Only export chats created on or before this date. Format: YYYY-MM-DD",
)
parser.add_argument(
"--user-id... | Could we use `action="store_true"` for these filters? Would just keep it consistent with other boolean args for this and other scripts | type=parse_date,
help="Only export chats created on or after this date. Format: YYYY-MM-DD",
)
parser.add_argument(
"--to-date",
type=parse_date,
help="Only export chats created on or before this date. Format: YYYY-MM-DD",
)
parser.add_argument(
"--user-id... | @@ -219,19 +226,68 @@ def parse_args() -> argparse.Namespace:
type=int,
help="Seed for the anonymizer. If not specified, no anonymization will be performed.",
)
- # TODO: filters: reported, score, user ID, chat ID, etc
- # TODO: date range?
+ parser.add_argument(
+ "--from-date",
... | inference/server/export.py | 26 | Python | 0.571 | suggestion | 134 | 51 | 51 | false | Improve inference export script | 3,071 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | simonusher | Review this pull request change.
Language: Python
File: inference/server/export.py
PR Title: Improve inference export script
BEFORE CODE: type=parse_date,
help="Only export chats created on or after this date. Format: YYYY-MM-DD",
)
parser.add_argument(
"--to-date",
type=pars... | Could we use `action="store_true"` for these filters? Would just keep it consistent with other boolean args for this and other scripts | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
gap="1"
height="full"
minH="0"
display="flex"
flexDirection="column"
bg="whiteAlpha.400"
_dark={{
bg: "blackAlpha.400",
}}
{...props}
>
<Flex>
<CreateChatButton
py="5"
leftIcon={<Plus size="16px" />}
variant="o... | Can you make it a selection instead? So user can select visible chat, hidden chat, trainable chat, and maybe shared chat in the future | flexDirection="column"
bg="whiteAlpha.400"
_dark={{
bg: "blackAlpha.400",
}}
{...props}
>
<Flex flexDirection={["column", "row"]} alignItems="stretch" p="2" gap="3">
<CreateChatButton
leftIcon={<Plus size="16px" />}
variant="outline"
... | @@ -78,19 +80,22 @@ export const ChatListBase = memo(function ChatListBase({
}}
{...props}
>
- <CreateChatButton
- py="5"
- leftIcon={<Plus size="16px"></Plus>}
- variant="outline"
- justifyContent="start"
- colorScheme="blue"
- borderRadius="lg"
- ... | website/src/components/Chat/ChatListBase.tsx | 26 | TypeScript | 0.429 | suggestion | 134 | 49 | 45 | false | Show hidden chats in chat list | 3,075 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatListBase.tsx
PR Title: Show hidden chats in chat list
BEFORE CODE: gap="1"
height="full"
minH="0"
display="flex"
flexDirection="column"
bg="whiteAlpha.400"
_dark={{
bg: "blac... | Can you make it a selection instead? So user can select visible chat, hidden chat, trainable chat, and maybe shared chat in the future | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import { Box } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { memo, ReactNode } from "react";
import SimpleBar from "simplebar-react";
export const ChatConfigDesktop = memo(function ChatConfigDesktop({ children }: { children: ReactNode }) {
const { t } = useTranslation("chat");
re... | if min and max are the same, can we just use `w`? | import { Box } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { memo, ReactNode } from "react";
export const ChatConfigDesktop = memo(function ChatConfigDesktop({ children }: { children: ReactNode }) {
const { t } = useTranslation("chat");
return (
<Box
py="4"
pl="4"... | @@ -14,6 +13,7 @@ export const ChatConfigDesktop = memo(function ChatConfigDesktop({ children }: {
minH="0"
flexDirection="column"
minW="270px"
+ maxW="270px" | website/src/components/Chat/ChatConfigDesktop.tsx | 16 | TypeScript | 0.286 | question | 49 | 37 | 29 | false | Allow to save custom preset | 3,079 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConfigDesktop.tsx
PR Title: Allow to save custom preset
BEFORE CODE: import { Box } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { memo, ReactNode } from "react";
import SimpleBar from "simp... | if min and max are the same, can we just use `w`? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
"""
@managed_tx_function(CommitMode.COMMIT)
def emoji_tx(session: deps.Session):
pr = PromptRepository(session, api_client, client_user=request.user)
return pr.handle_message_emoji(message_id, request.op, request.emoji)
return utils.prepare_message(emoji_tx())
@router.get("/search", ... | Can you merge this to the /cursor endpoint? | from datetime import datetime
from typing import Optional
from uuid import UUID
from fastapi import APIRouter, Depends, Query
from loguru import logger
from oasst_backend.api import deps
from oasst_backend.api.v1 import utils
from oasst_backend.models import ApiClient, MessageTreeState
from oasst_backend.prompt_reposi... | @@ -354,3 +354,21 @@ def emoji_tx(session: deps.Session):
return pr.handle_message_emoji(message_id, request.op, request.emoji)
return utils.prepare_message(emoji_tx())
+
+
+@router.get("/search", response_model=list[protocol.Message])
+def search_messages(
+ *,
+ query: str,
+ lang: str,
+ ... | backend/oasst_backend/api/v1/messages.py | 26 | Python | 0.071 | question | 43 | 27 | 359 | false | Backend support for message search | 3,080 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | olliestanley | Review this pull request change.
Language: Python
File: backend/oasst_backend/api/v1/messages.py
PR Title: Backend support for message search
BEFORE CODE: """
@managed_tx_function(CommitMode.COMMIT)
def emoji_tx(session: deps.Session):
pr = PromptRepository(session, api_client, client_user=requ... | Can you merge this to the /cursor endpoint? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
@app.task(name="hf_feature_extraction")
def hf_feature_extraction(text, message_id, api_client):
try:
with default_session_factory() as session:
model_name: str = HfEmbeddingModel.MINILM.value
url: str = HfUrl.HUGGINGFACE_FEATURE_EXTRACTION.value
embedding = async_to_sync... | Should we limit the messages returned from db here? On the first run, it will load about 160k messages at once, which might break the server | @app.task(name="hf_feature_extraction")
def hf_feature_extraction(text, message_id, api_client):
try:
with default_session_factory() as session:
model_name: str = HfEmbeddingModel.MINILM.value
url: str = HfUrl.HUGGINGFACE_FEATURE_EXTRACTION.value
embedding = async_to_sync... | @@ -67,6 +69,23 @@ def hf_feature_extraction(text, message_id, api_client):
logger.error(f"Could not extract embedding for text reply to {message_id=} with {text=} by.error {str(e)}")
+@shared_task(name="update_search_vectors")
+def update_search_vectors(batch_size: int) -> None:
+ logger.info("update_s... | backend/oasst_backend/scheduled_tasks.py | 26 | Python | 0.429 | question | 140 | 51 | 51 | false | Backend support for message search | 3,080 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | olliestanley | Review this pull request change.
Language: Python
File: backend/oasst_backend/scheduled_tasks.py
PR Title: Backend support for message search
BEFORE CODE: @app.task(name="hf_feature_extraction")
def hf_feature_extraction(text, message_id, api_client):
try:
with default_session_factory() as session:
... | Should we limit the messages returned from db here? On the first run, it will load about 160k messages at once, which might break the server | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
from enum import IntEnum
from functools import wraps
from http import HTTPStatus
from typing import Callable
from loguru import logger
from oasst_backend.config import settings
from oasst_backend.database import engine
from oasst_shared.exceptions import OasstError, OasstErrorCode
from psycopg2.errors import DeadlockD... | The front end recently has a few new translation langs, can you check it? | from enum import IntEnum
from functools import wraps
from http import HTTPStatus
from typing import Callable
from loguru import logger
from oasst_backend.config import settings
from oasst_backend.database import engine
from oasst_shared.exceptions import OasstError, OasstErrorCode
from psycopg2.errors import DeadlockD... | @@ -15,6 +15,37 @@
Error Handling Reference: https://www.postgresql.org/docs/15/mvcc-serialization-failure-handling.html
"""
+TEXT_SEARCH_LANGUAGE_MAPPING: dict[str, str] = { | backend/oasst_backend/utils/database_utils.py | 18 | Python | 0.214 | question | 73 | 43 | 43 | false | Backend support for message search | 3,080 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | olliestanley | Review this pull request change.
Language: Python
File: backend/oasst_backend/utils/database_utils.py
PR Title: Backend support for message search
BEFORE CODE: from enum import IntEnum
from functools import wraps
from http import HTTPStatus
from typing import Callable
from loguru import logger
from oasst_backend.c... | The front end recently has a few new translation langs, can you check it? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
FlaggedMessage,
Message,
MessageEmbedding,
MessageEmoji,
MessageReaction,
MessageToxicity,
MessageTreeState,
Task,
TextLabels,
User,
message_tree_state,
)
from oasst_backend.models.payload_column_type import PayloadContainer
from oasst_backend.task_repository import TaskRepos... | I don't think we should have random TODOs like this, if we want to embark on adding docstrings throughout the code we should raise a GitHub issue for it instead | FlaggedMessage,
Message,
MessageEmbedding,
MessageEmoji,
MessageReaction,
MessageRevision,
MessageToxicity,
MessageTreeState,
Task,
TextLabels,
User,
message_tree_state,
)
from oasst_backend.models.payload_column_type import PayloadContainer
from oasst_backend.task_reposi... | @@ -39,6 +39,9 @@
from sqlalchemy.orm.attributes import flag_modified
from sqlmodel import JSON, Session, and_, func, literal_column, not_, or_, text, update
+# TODO: Add doc string to all methods and classes. | backend/oasst_backend/prompt_repository.py | 26 | Python | 0.429 | suggestion | 160 | 51 | 51 | false | Make send report message from discord a celery task. | 3,089 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | espoirMur | Review this pull request change.
Language: Python
File: backend/oasst_backend/prompt_repository.py
PR Title: Make send report message from discord a celery task.
BEFORE CODE: FlaggedMessage,
Message,
MessageEmbedding,
MessageEmoji,
MessageReaction,
MessageToxicity,
MessageTreeState,
... | I don't think we should have random TODOs like this, if we want to embark on adding docstrings throughout the code we should raise a GitHub issue for it instead | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
from __future__ import absolute_import, unicode_literals
from datetime import datetime
from typing import Any, Dict, List
from asgiref.sync import async_to_sync
from celery import shared_task
from loguru import logger
from oasst_backend.celery_worker import app as celery_app
from oasst_backend.models import ApiClient... | I also don't think this should be a TODO, for now we have very few tasks so one file is file and probably will be for a long time | from __future__ import absolute_import, unicode_literals
from datetime import timedelta
from typing import Any, Dict, List
from asgiref.sync import async_to_sync
from celery import shared_task
from loguru import logger
from oasst_backend.celery_worker import app
from oasst_backend.models import ApiClient, Message, Us... | @@ -15,6 +15,8 @@
from oasst_shared.utils import utcnow
from sqlmodel import select
+# TODO: move each task to a separate file in the future to avoid cluttering this file | backend/oasst_backend/scheduled_tasks.py | 18 | Python | 0.429 | suggestion | 129 | 43 | 43 | false | Make send report message from discord a celery task. | 3,089 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | espoirMur | Review this pull request change.
Language: Python
File: backend/oasst_backend/scheduled_tasks.py
PR Title: Make send report message from discord a celery task.
BEFORE CODE: from __future__ import absolute_import, unicode_literals
from datetime import datetime
from typing import Any, Dict, List
from asgiref.sync i... | I also don't think this should be a TODO, for now we have very few tasks so one file is file and probably will be for a long time | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
OasstErrorCode.TEXT_LABELS_DUPLICATE_TASK_REPLY,
)
message.review_count += 1
self.db.add(message)
self.db.add(model)
return model, task, message
def fetch_random_message_tree(
self,
require_role: str = Non... | So I think what we want to do here is add this to a queue (maybe in the backend Redis) and then run the Celery task on a timer, which will send all recently flagged messages as a single Discord message, this way we can avoid spamming Discord with API calls on every flag | if not task.collective:
task.done = True
self.db.add(task)
logger.debug(f"inserting TextLabels for {message_id=}, {text_labels_id=}")
model = TextLabels(
id=text_labels_id,
api_client_id=self.api_client.id,
message_id=messa... | @@ -547,7 +549,7 @@ def store_text_labels(self, text_labels: protocol_schema.TextLabels) -> tuple[Te
message_id, protocol_schema.EmojiOp.add, protocol_schema.EmojiCode.red_flag
)
- discord.send_new_report_message(message=message, label_text=text_labels.... | backend/oasst_backend/prompt_repository.py | 26 | Python | 0.5 | suggestion | 270 | 51 | 51 | false | Make send report message from discord a celery task. | 3,089 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | espoirMur | Review this pull request change.
Language: Python
File: backend/oasst_backend/prompt_repository.py
PR Title: Make send report message from discord a celery task.
BEFORE CODE: OasstErrorCode.TEXT_LABELS_DUPLICATE_TASK_REPLY,
)
message.review_count += 1
... | So I think what we want to do here is add this to a queue (maybe in the backend Redis) and then run the Celery task on a timer, which will send all recently flagged messages as a single Discord message, this way we can avoid spamming Discord with API calls on every flag | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
logger.warning(f"JSON is still not valid, trying to fix it with LLM {fixed_json}")
# if it's still not valid, try with LLM fixer
prompt = f"""{V2_PROMPTER_PREFIX}Below is malformed JSON object string:
--------------
{json_str}
--------------
Parsing error:
--------------
{e}
RULES:
... | This code could just be something like:
```
best_match, best_similarity = max(
((tool, similarity(tool.name, tool_name)) for tool in tools),
key=lambda x: x[1],
default=(None, 0),
)
``` | logger.warning(f"JSON is still not valid, trying to fix it with LLM {fixed_json}")
# if it's still not valid, try with LLM fixer
prompt = f"""{V2_PROMPTER_PREFIX}Below is malformed JSON object string:
--------------
{json_str}
--------------
Parsing error:
--------------
{e}
RULES:
... | @@ -162,12 +162,17 @@ def prepare_json(json_str: str) -> str:
def use_tool(tool_name: str, tool_input: str, tools: list) -> str:
+ best_match = None | inference/worker/chat_chain_utils.py | 26 | Python | 0.857 | suggestion | 227 | 51 | 51 | false | Use the tool with the highest similarity instead of the first one over 75% | 3,100 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | someone13574 | Review this pull request change.
Language: Python
File: inference/worker/chat_chain_utils.py
PR Title: Use the tool with the highest similarity instead of the first one over 75%
BEFORE CODE: logger.warning(f"JSON is still not valid, trying to fix it with LLM {fixed_json}")
# if it's still no... | This code could just be something like:
```
best_match, best_similarity = max(
((tool, similarity(tool.name, tool_name)) for tool in tools),
key=lambda x: x[1],
default=(None, 0),
)
``` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
const useAutoScroll = (messages: InferenceMessage[], streamedResponse: string | null) => {
const enableAutoScroll = useRef(true);
const messagesEndRef = useRef<HTMLDivElement>(null);
const chatContainerRef = useRef<HTMLDivElement>(null);
const updateEnableAutoScroll = useCallback(() => {
const container = c... | Does this happen on submit? or does it run always? | <ChatWarning />
</Box>
);
});
const useAutoScroll = (messages: InferenceMessage[], streamedResponse: string | null) => {
const enableAutoScroll = useRef(true);
const messagesEndRef = useRef<HTMLDivElement>(null);
const chatContainerRef = useRef<HTMLDivElement>(null);
const updateEnableAutoScroll = ... | @@ -327,7 +327,12 @@ const useAutoScroll = (messages: InferenceMessage[], streamedResponse: string |
}
messagesEndRef.current?.scrollIntoView({ behavior: "smooth" });
- }, [messages, streamedResponse]);
+ }, [streamedResponse]);
+
+ useEffect(() => {
+ messagesEndRef.current?.scrollIntoView({ behavior... | website/src/components/Chat/ChatConversation.tsx | 26 | TypeScript | 0.214 | question | 50 | 46 | 51 | false | Scroll to end of chat on submit | 3,189 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConversation.tsx
PR Title: Scroll to end of chat on submit
BEFORE CODE: const useAutoScroll = (messages: InferenceMessage[], streamedResponse: string | null) => {
const enableAutoScroll = useRef(true);
const messagesEndR... | Does this happen on submit? or does it run always? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
error_message: str | None = None
status: Literal["success", "failure"]
class PluginUsed(pydantic.BaseModel):
name: str | None = None
url: str | None = None
trusted: bool | None = None
execution_details: PluginExecutionDetails
def make_seed() -> int:
return random.randint(0, 0xFFFF_FFFF_F... | ```suggestion
plugin_max_depth: int = 4
```
maybe use the same default everywhere? | error_message: str | None = None
status: Literal["success", "failure"]
class PluginUsed(pydantic.BaseModel):
name: str | None = None
url: str | None = None
trusted: bool | None = None
execution_details: PluginExecutionDetails
def make_seed() -> int:
return random.randint(0, 0xFFFF_FFFF_F... | @@ -201,6 +201,7 @@ class WorkParameters(pydantic.BaseModel):
default_factory=make_seed,
)
plugins: list[PluginEntry] = pydantic.Field(default_factory=list[PluginEntry])
+ plugin_max_depth: int = 6 | oasst-shared/oasst_shared/schemas/inference.py | 26 | Python | 0.643 | question | 89 | 51 | 51 | false | Add inference server setting to control max depth of plugins | 3,227 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | olliestanley | Review this pull request change.
Language: Python
File: oasst-shared/oasst_shared/schemas/inference.py
PR Title: Add inference server setting to control max depth of plugins
BEFORE CODE: error_message: str | None = None
status: Literal["success", "failure"]
class PluginUsed(pydantic.BaseModel):
name: ... | ```suggestion
plugin_max_depth: int = 4
```
maybe use the same default everywhere? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
await handle_generated_text_response(
work_request_map=work_request_map,
response=worker_response,
)
await _update_session(worker_response.metrics)
... | This should probably just be "plugin_intermediate" to match others | await handle_generated_text_response(
work_request_map=work_request_map,
response=worker_response,
)
await _update_session(worker_response.metrics)
... | @@ -218,6 +218,12 @@ def _add_receive(ftrs: set):
response=worker_response,
work_request_map=work_request_map,
)
+ case "plugin_intermediate_response": | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.286 | suggestion | 66 | 51 | 51 | false | Plugins intermediate steps reporting | 3,238 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | draganjovanovich | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Plugins intermediate steps reporting
BEFORE CODE: await handle_generated_text_response(
work_request_map=work_request_map,
... | This should probably just be "plugin_intermediate" to match others | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
inner_monologue.append("In: " + str(init_prompt))
inner_monologue.append("Out: " + str(chain_response))
current_action_thought = ""
if THOUGHT_SEQ in chain_response:
current_action_thought = chain_response.split(THOUGHT_SEQ)[1].split("\n")[0]
# Tool name/assistant prefix, Tool input/assist... | This seems to have been a bad merge with main - `MAX_DEPTH` constant doesn't exist anymore, should be using the `plugin_max_depth` parameter | inner_monologue.append("In: " + str(init_prompt))
inner_monologue.append("Out: " + str(chain_response))
current_action_thought = ""
if THOUGHT_SEQ in chain_response:
current_action_thought = chain_response.split(THOUGHT_SEQ)[1].split("\n")[0]
# Tool name/assistant prefix, Tool input/assist... | @@ -142,18 +145,47 @@ def handle_plugin_usage(
tokenizer, worker_config, parameters, prompt_template, memory, tool_names, language, action_input_format
)
+ # send "thinking..." intermediate step to UI (This will discard queue position 0) immediately
+ utils.send_response(
+ ws,
+ inf... | inference/worker/chat_chain.py | 26 | Python | 0.571 | suggestion | 140 | 51 | 51 | false | Plugins intermediate steps reporting | 3,238 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | draganjovanovich | Review this pull request change.
Language: Python
File: inference/worker/chat_chain.py
PR Title: Plugins intermediate steps reporting
BEFORE CODE: inner_monologue.append("In: " + str(init_prompt))
inner_monologue.append("Out: " + str(chain_response))
current_action_thought = ""
if THOUGHT_SEQ in ch... | This seems to have been a bad merge with main - `MAX_DEPTH` constant doesn't exist anymore, should be using the `plugin_max_depth` parameter | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
await onToken(tokens);
} else if (chunk.event_type === "message") {
// final message
return chunk.message;
} else if (chunk.event_type === "error") {
// handle error
await onError(chunk.error);
return chunk.message;
} else {
console.error("Unexpe... | Why do we need to change to log? I think error will make more sense here | } else if (event === "ping") {
continue;
}
try {
const chunk: InferenceEvent = JSON.parse(data);
if (chunk.event_type === "pending") {
await onPending({ queuePosition: chunk.queue_position, queueSize: chunk.queue_size });
} else if (chunk.event_type === "token") {
tok... | @@ -39,8 +48,15 @@ export async function handleChatEventStream({
// handle error
await onError(chunk.error);
return chunk.message;
+ } else if (chunk.event_type === "plugin_intermediate") {
+ await onPluginIntermediateResponse({
+ currentPluginThought: chunk.current_plugi... | website/src/lib/chat_stream.ts | 26 | TypeScript | 0.429 | question | 72 | 51 | 51 | false | Plugins intermediate steps reporting | 3,238 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | draganjovanovich | Review this pull request change.
Language: TypeScript
File: website/src/lib/chat_stream.ts
PR Title: Plugins intermediate steps reporting
BEFORE CODE: await onToken(tokens);
} else if (chunk.event_type === "message") {
// final message
return chunk.message;
} else if (chunk.event... | Why do we need to change to log? I think error will make more sense here | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
const messagesEndRef = useRef<HTMLDivElement>(null);
const chatContainerRef = useRef<HTMLDivElement>(null);
const updateEnableAutoScroll = useCallback(() => {
const container = chatContainerRef.current;
if (!container) {
return;
}
const isEnable = container.scrollHeight - container.scrollTo... | Have you check UI in the light mode? | <Box
position="absolute"
bottom="0"
left="50%"
transform="translate(-50%)"
display="flex"
flexDirection="row"
gap="1"
justifyContent="center"
alignItems="center"
>
<Box
... | @@ -305,6 +313,33 @@ export const ChatConversation = memo(function ChatConversation({ chatId, getConf
{t("queue_info", queueInfo)}
</Badge>
)}
+ {pluginIntermediateResponse && pluginIntermediateResponse.currentPluginThought && (
+ <Box
+ position="absolute"
+ ... | website/src/components/Chat/ChatConversation.tsx | 26 | TypeScript | 0.071 | question | 36 | 44 | 51 | false | Plugins intermediate steps reporting | 3,238 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | draganjovanovich | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConversation.tsx
PR Title: Plugins intermediate steps reporting
BEFORE CODE: const messagesEndRef = useRef<HTMLDivElement>(null);
const chatContainerRef = useRef<HTMLDivElement>(null);
const updateEnableAutoScroll = us... | Have you check UI in the light mode? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
return limited_images
def truncate_paragraphs(paragraphs, max_length):
truncated_paragraphs = []
current_length = 0
for paragraph in paragraphs:
if current_length + len(paragraph) <= max_length:
truncated_paragraphs.append(paragraph)
current_length += len(paragraph)
... | Is there a reason we need to use `httpx` over `aiohttp` here? The server already depends on `aiohttp` so we could reduce dependencies by using it, if possible | current_length += url_length
return limited_images
def truncate_paragraphs(paragraphs, max_length):
truncated_paragraphs = []
current_length = 0
for paragraph in paragraphs:
if len(paragraph) == 0:
continue
paragraph = paragraph.strip()
if current_length +... | @@ -0,0 +1,253 @@
+import codecs
+import io
+import json
+import re
+from pathlib import Path
+
+import httpx
+import PyPDF2
+import yaml
+from bs4 import BeautifulSoup
+from fastapi import FastAPI, Query, Request, Response
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.openapi.utils import get_opena... | inference/server/oasst_inference_server/plugins/web_retriever/main.py | 26 | Python | 0.571 | suggestion | 158 | 51 | 51 | false | Add Dragan's web retriever plugin | 3,277 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | andreaskoepf | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/plugins/web_retriever/main.py
PR Title: Add Dragan's web retriever plugin
BEFORE CODE: return limited_images
def truncate_paragraphs(paragraphs, max_length):
truncated_paragraphs = []
current_length = 0
... | Is there a reason we need to use `httpx` over `aiohttp` here? The server already depends on `aiohttp` so we could reduce dependencies by using it, if possible | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
text_yaml = f"text_content: {MULTILINE_SYM}\n"
for line in text.split("\n"):
text_yaml += f" {line}\n"
images_yaml = "images:\n" if len(images) > 0 else ""
for image in images:
images_yaml += f"- {image}\n"
yaml_text = f"{text_yaml}\n{images_yaml}"
... | would it be simpler to:
```py
from starlette.responses import FileResponse
@app.get("/icon.png", include_in_schema=False)
async def api_icon():
return FileResponse(folder_path / "icon.png")
``` | error_message = f"Sorry, unsupported content type '{content_type}' at {url}.\nYou should report this message to the user!"
return JSONResponse(content={"error": error_message}, status_code=500)
images = [f"http:{url}" if url.startswith("//") else url for url in images]
images = ... | @@ -0,0 +1,253 @@
+import codecs
+import io
+import json
+import re
+from pathlib import Path
+
+import httpx
+import PyPDF2
+import yaml
+from bs4 import BeautifulSoup
+from fastapi import FastAPI, Query, Request, Response
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.openapi.utils import get_opena... | inference/server/oasst_inference_server/plugins/web_retriever/main.py | 26 | Python | 0.786 | suggestion | 208 | 51 | 51 | false | Add Dragan's web retriever plugin | 3,277 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | andreaskoepf | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/plugins/web_retriever/main.py
PR Title: Add Dragan's web retriever plugin
BEFORE CODE: text_yaml = f"text_content: {MULTILINE_SYM}\n"
for line in text.split("\n"):
text_yaml += f" {line}\n"
... | would it be simpler to:
```py
from starlette.responses import FileResponse
@app.get("/icon.png", include_in_schema=False)
async def api_icon():
return FileResponse(folder_path / "icon.png")
``` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
@router.delete("/{message_id}", status_code=HTTP_204_NO_CONTENT)
def mark_message_deleted(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: ApiClient = Depends(deps.get_trusted_api_client),
db: Session = Depends(deps.get_db),
):
pr = Prom... | nit: I think this should be 200, since it is processed immediately. |
@router.delete("/{message_id}", status_code=HTTP_204_NO_CONTENT)
def mark_message_deleted(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: ApiClient = Depends(deps.get_trusted_api_client),
db: Session = Depends(deps.get_db),
):
pr = Prom... | @@ -339,6 +339,18 @@ def undelete_message(
pr.undelete_deleted_message(message_id)
+@router.post("/{message_id}/edit", status_code=HTTP_202_ACCEPTED) | backend/oasst_backend/api/v1/messages.py | 26 | Python | 0.286 | suggestion | 67 | 51 | 51 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: Python
File: backend/oasst_backend/api/v1/messages.py
PR Title: Moderator Message Editing
BEFORE CODE:
@router.delete("/{message_id}", status_code=HTTP_204_NO_CONTENT)
def mark_message_deleted(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(... | nit: I think this should be 200, since it is processed immediately. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
from datetime import datetime
from typing import Optional
from uuid import UUID, uuid4
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from sqlmodel import Field, SQLModel
from .payload_column_type import PayloadContainer, payload_column_type
class MessageRevision(SQLModel, table=True):
__ta... | would it be possible to use uuid7? | from datetime import datetime
from typing import Optional
from uuid import UUID
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from sqlmodel import Field, SQLModel
from uuid_extensions import uuid7
from .payload_column_type import PayloadContainer, payload_column_type
class MessageRevision(SQLM... | @@ -0,0 +1,31 @@
+from datetime import datetime
+from typing import Optional
+from uuid import UUID, uuid4 | backend/oasst_backend/models/message_revision.py | 3 | Python | 0.143 | question | 34 | 28 | 26 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: Python
File: backend/oasst_backend/models/message_revision.py
PR Title: Moderator Message Editing
BEFORE CODE: from datetime import datetime
from typing import Optional
from uuid import UUID, uuid4
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from s... | would it be possible to use uuid7? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import sqlalchemy.dialects.postgresql as pg
from sqlmodel import Field, SQLModel
from .payload_column_type import PayloadContainer, payload_column_type
class MessageRevision(SQLModel, table=True):
__tablename__ = "message_revision"
id: UUID = Field(
sa_column=sa.Column(
pg.UUID(as_uuid=T... | could this be null if there are no previous revisions? | from datetime import datetime
from typing import Optional
from uuid import UUID
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from sqlmodel import Field, SQLModel
from uuid_extensions import uuid7
from .payload_column_type import PayloadContainer, payload_column_type
class MessageRevision(SQLM... | @@ -0,0 +1,31 @@
+from datetime import datetime
+from typing import Optional
+from uuid import UUID, uuid4
+
+import sqlalchemy as sa
+import sqlalchemy.dialects.postgresql as pg
+from sqlmodel import Field, SQLModel
+
+from .payload_column_type import PayloadContainer, payload_column_type
+
+
+class MessageRevision(SQ... | backend/oasst_backend/models/message_revision.py | 26 | Python | 0.429 | question | 54 | 27 | 26 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: Python
File: backend/oasst_backend/models/message_revision.py
PR Title: Moderator Message Editing
BEFORE CODE: import sqlalchemy.dialects.postgresql as pg
from sqlmodel import Field, SQLModel
from .payload_column_type import PayloadContainer, payload_column_type
class M... | could this be null if there are no previous revisions? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
cell: ({ getValue, row }) => {
const limit = 95;
const text = getValue();
const isActive = row.original.isActive;
const renderText = isActive ? text : text.length > limit ? `${text.slice(0, limit)}...` : text;
return (
<Box wordBreak="break-all" whi... | can you use `marginStart` or `ms` instead of `ml`? for RTL | cell: ({ getValue, row }) => {
const limit = 95;
const text = getValue();
const isActive = row.original.isActive;
const renderText = isActive ? text : text.length > limit ? `${text.slice(0, limit)}...` : text;
return (
<Box wordBreak="break-all" whi... | @@ -150,7 +150,16 @@ export const AdminMessageTable = ({ userId, includeUser }: { userId?: string; in
Deleted
</Badge>
)}
- {row.original.review_result === false && <Badge colorScheme="yellow">Spam</Badge>}
+ {row.original.review_result === fa... | website/src/components/Messages/AdminMessageTable.tsx | 26 | TypeScript | 0.429 | suggestion | 58 | 51 | 51 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/AdminMessageTable.tsx
PR Title: Moderator Message Editing
BEFORE CODE: cell: ({ getValue, row }) => {
const limit = 95;
const text = getValue();
const isActive = row.original.isActive;
... | can you use `marginStart` or `ms` instead of `ml`? for RTL | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
const router = useRouter();
const [messageText, setMessageText] = useState("");
const [submitting, setSubmitting] = useBoolean(false);
const { data, isLoading, error } = useSWRImmutable<{ tree: MessageWithChildren | null; message?: Message }>(
`/api/admin/messages/${id}/tree`,
get,
{ keepPreviousDa... | please `useCallback` on all of these functions | const { t } = useTranslation(["common", "tasks"]);
const router = useRouter();
const [messageText, setMessageText] = useState("");
const { trigger: submitEdit, isMutating: submitting } = useSWRMutation(API_ROUTES.ADMIN_EDIT_MESSAGE(id), post, {
onSuccess: () => {
router.push(ROUTES.ADMIN_MESSAGE_DETA... | @@ -0,0 +1,176 @@
+import {
+ Box,
+ Button,
+ Card,
+ CardBody,
+ Flex,
+ Stack,
+ Tab,
+ TabList,
+ TabPanel,
+ TabPanels,
+ Tabs,
+ Text,
+ useColorModeValue,
+} from "@chakra-ui/react";
+import { GetServerSideProps, InferGetServerSidePropsType } from "next";
+import Head from "next/head";
+import { use... | website/src/pages/admin/edit/[id].tsx | 26 | TypeScript | 0.286 | suggestion | 46 | 51 | 51 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: TypeScript
File: website/src/pages/admin/edit/[id].tsx
PR Title: Moderator Message Editing
BEFORE CODE: const router = useRouter();
const [messageText, setMessageText] = useState("");
const [submitting, setSubmitting] = useBoolean(false);
const { data, isLoading, ... | please `useCallback` on all of these functions | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
() => (
<Suspense fallback={messageText}>
<RenderedMarkdown markdown={messageText}></RenderedMarkdown>
</Suspense>
),
[messageText]
);
const getThreadFromTree = (tree: MessageWithChildren | null, targetMessageId: string) => {
if (!tree) {
return null;
}
if (tree.i... | would it be an option to use swr mutations?
https://swr.vercel.app/docs/mutation#useswrmutation
This would also eliminate the need for `isSubmitting` | setMessageText(data.message.text);
}
}, [data]);
const previewContent = useMemo(
() => (
<Suspense fallback={messageText}>
<RenderedMarkdown markdown={messageText}></RenderedMarkdown>
</Suspense>
),
[messageText]
);
const getThreadFromTree = useCallback((tree: Message... | @@ -0,0 +1,176 @@
+import {
+ Box,
+ Button,
+ Card,
+ CardBody,
+ Flex,
+ Stack,
+ Tab,
+ TabList,
+ TabPanel,
+ TabPanels,
+ Tabs,
+ Text,
+ useColorModeValue,
+} from "@chakra-ui/react";
+import { GetServerSideProps, InferGetServerSidePropsType } from "next";
+import Head from "next/head";
+import { use... | website/src/pages/admin/edit/[id].tsx | 26 | TypeScript | 0.571 | question | 155 | 51 | 51 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: TypeScript
File: website/src/pages/admin/edit/[id].tsx
PR Title: Moderator Message Editing
BEFORE CODE: () => (
<Suspense fallback={messageText}>
<RenderedMarkdown markdown={messageText}></RenderedMarkdown>
</Suspense>
),
[messageText]
);
... | would it be an option to use swr mutations?
https://swr.vercel.app/docs/mutation#useswrmutation
This would also eliminate the need for `isSubmitting` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
onClick={handleSubmit}
isLoading={submitting}
loadingText={"Submitting"}
>
{t("common:submit")}
</Button>
</Flex>
</Stack>
</>
</TwoColumnsWithCards>
... | no need to define the individual namespaces here, `await serverSideTranslations(locale)` is good enough. | isLoading={submitting}
loadingText={"Submitting"}
>
{t("common:submit")}
</Button>
</Flex>
</Stack>
</>
</TwoColumnsWithCards>
</AdminArea>
)}
</>
);
};... | @@ -0,0 +1,176 @@
+import {
+ Box,
+ Button,
+ Card,
+ CardBody,
+ Flex,
+ Stack,
+ Tab,
+ TabList,
+ TabPanel,
+ TabPanels,
+ Tabs,
+ Text,
+ useColorModeValue,
+} from "@chakra-ui/react";
+import { GetServerSideProps, InferGetServerSidePropsType } from "next";
+import Head from "next/head";
+import { use... | website/src/pages/admin/edit/[id].tsx | 26 | TypeScript | 0.5 | suggestion | 104 | 31 | 30 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | someone13574 | Review this pull request change.
Language: TypeScript
File: website/src/pages/admin/edit/[id].tsx
PR Title: Moderator Message Editing
BEFORE CODE: onClick={handleSubmit}
isLoading={submitting}
loadingText={"Submitting"}
>
... | no need to define the individual namespaces here, `await serverSideTranslations(locale)` is good enough. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
@router.delete("/{message_id}", status_code=HTTP_204_NO_CONTENT)
def mark_message_deleted(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: ApiClient = Depends(deps.get_trusted_api_client),
db: Session = Depends(deps.get_db),
):
pr = Prom... | I don't think `HTTP_200_OK` needs specifying |
@router.delete("/{message_id}", status_code=HTTP_204_NO_CONTENT)
def mark_message_deleted(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: ApiClient = Depends(deps.get_trusted_api_client),
db: Session = Depends(deps.get_db),
):
pr = Prom... | @@ -339,6 +339,18 @@ def undelete_message(
pr.undelete_deleted_message(message_id)
+@router.post("/{message_id}/edit", status_code=HTTP_200_OK) | backend/oasst_backend/api/v1/messages.py | 26 | Python | 0.214 | suggestion | 44 | 51 | 51 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | someone13574 | Review this pull request change.
Language: Python
File: backend/oasst_backend/api/v1/messages.py
PR Title: Moderator Message Editing
BEFORE CODE:
@router.delete("/{message_id}", status_code=HTTP_204_NO_CONTENT)
def mark_message_deleted(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(... | I don't think `HTTP_200_OK` needs specifying | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
from datetime import datetime
from typing import Optional
from uuid import UUID
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from sqlmodel import Field, SQLModel
from uuid_extensions import uuid7
from .payload_column_type import PayloadContainer, payload_column_type
class MessageRevision(SQLM... | Why is this `Optional` even though it is not `nullable`? | from datetime import datetime
from typing import Optional
from uuid import UUID
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from sqlmodel import Field, SQLModel
from uuid_extensions import uuid7
from .payload_column_type import PayloadContainer, payload_column_type
class MessageRevision(SQLM... | @@ -0,0 +1,25 @@
+from datetime import datetime
+from typing import Optional
+from uuid import UUID
+
+import sqlalchemy as sa
+import sqlalchemy.dialects.postgresql as pg
+from sqlmodel import Field, SQLModel
+from uuid_extensions import uuid7
+
+from .payload_column_type import PayloadContainer, payload_column_type
+... | backend/oasst_backend/models/message_revision.py | 23 | Python | 0.5 | question | 56 | 26 | 26 | false | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | someone13574 | Review this pull request change.
Language: Python
File: backend/oasst_backend/models/message_revision.py
PR Title: Moderator Message Editing
BEFORE CODE: from datetime import datetime
from typing import Optional
from uuid import UUID
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from sqlmodel... | Why is this `Optional` even though it is not `nullable`? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
"payload": payload,
}
if "tags" in details:
tag_name = details["tags"][0]
endpoint_data["tag"] = tag_name
endpoint = inference.PluginOpenAPIEndpoint(**endpoint_data)
return endpoint
def get_plugin_endpoints(api_url: str, openapi_dict: dict) -> list[inference.PluginOpenAPIEndp... | Could we restore this? Feels a bit cleaner than an extra layer of nesting in the below code | "payload": payload,
}
if "tags" in details:
tag_name = details["tags"][0]
endpoint_data["tag"] = tag_name
endpoint = inference.PluginOpenAPIEndpoint(**endpoint_data)
return endpoint
def get_plugin_endpoints(api_url: str, openapi_dict: dict) -> list[inference.PluginOpenAPIEndp... | @@ -135,19 +140,20 @@ def get_plugin_endpoints(api_url: str, openapi_dict: dict) -> list[inference.Plu
def prepare_plugin_for_llm(plugin_url: str) -> inference.PluginConfig | None:
plugin_config = get_plugin_config(plugin_url)
- if not plugin_config: | inference/worker/openapi_parser.py | 26 | Python | 0.286 | suggestion | 91 | 48 | 49 | false | fix for plugins openapi parser | 3,316 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | draganjovanovich | Review this pull request change.
Language: Python
File: inference/worker/openapi_parser.py
PR Title: fix for plugins openapi parser
BEFORE CODE: "payload": payload,
}
if "tags" in details:
tag_name = details["tags"][0]
endpoint_data["tag"] = tag_name
endpoint = inference.Plugin... | Could we restore this? Feels a bit cleaner than an extra layer of nesting in the below code | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
message = self.db.query(Message).filter(Message.id == message_id).one_or_none()
if fail_if_missing and not message:
raise OasstError("Message not found", OasstErrorCode.MESSAGE_NOT_FOUND, HTTPStatus.NOT_FOUND)
return message
def fetch_non_task_text_labels(self, message_id: UUID... | Could you maybe add a comment here that we sort by the date part of the uuid7? Might not be directly obvious to everyone... |
message = self.db.query(Message).filter(Message.id == message_id).one_or_none()
if fail_if_missing and not message:
raise OasstError("Message not found", OasstErrorCode.MESSAGE_NOT_FOUND, HTTPStatus.NOT_FOUND)
return message
def fetch_non_task_text_labels(self, message_id: UUID... | @@ -800,6 +800,15 @@ def fetch_message_text_labels(self, message_id: UUID, user_id: Optional[UUID] =
query = query.filter(TextLabels.user_id == user_id)
return query.all()
+ def fetch_message_revision_history(self, message_id: UUID) -> list[MessageRevision]:
+ revisions: list[MessageRe... | backend/oasst_backend/prompt_repository.py | 26 | Python | 0.429 | question | 123 | 51 | 51 | false | Edit history viewer | 3,322 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | someone13574 | Review this pull request change.
Language: Python
File: backend/oasst_backend/prompt_repository.py
PR Title: Edit history viewer
BEFORE CODE:
message = self.db.query(Message).filter(Message.id == message_id).one_or_none()
if fail_if_missing and not message:
raise OasstError("Message not... | Could you maybe add a comment here that we sort by the date part of the uuid7? Might not be directly obvious to everyone... | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
export interface MessageHistoryTableProps {
message: Message;
revisions: MessageRevision[];
}
export function MessageHistoryTable({ message, revisions }: MessageHistoryTableProps) {
const { t } = useTranslation(["message"]);
const router = useRouter();
return (
<Stack spacing={4}>
{(revisions.leng... | You can use `Omit<MessageRevision, ...>` here to avoid lengthy code | export interface MessageHistoryTableProps {
message: Message;
revisions: MessageRevision[];
}
export function MessageHistoryTable({ message, revisions }: MessageHistoryTableProps) {
const { t } = useTranslation(["message"]);
const router = useRouter();
return (
<Stack spacing={4}>
{(revisions.leng... | @@ -0,0 +1,85 @@
+import { Badge, Flex, Stack, Tooltip } from "@chakra-ui/react";
+import { boolean } from "boolean";
+import { User } from "lucide-react";
+import { useRouter } from "next/router";
+import { useTranslation } from "next-i18next";
+import { ROUTES } from "src/lib/routes";
+import { Message, MessageRevisi... | website/src/components/Messages/MessageHistoryTable.tsx | 26 | TypeScript | 0.429 | suggestion | 67 | 51 | 51 | false | Edit history viewer | 3,322 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | someone13574 | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/MessageHistoryTable.tsx
PR Title: Edit history viewer
BEFORE CODE: export interface MessageHistoryTableProps {
message: Message;
revisions: MessageRevision[];
}
export function MessageHistoryTable({ message, revisions }... | You can use `Omit<MessageRevision, ...>` here to avoid lengthy code | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
}),
})) || []),
],
false
);
},
[mutateChatResponses]
);
const handleCreateChat = useCallback(() => {
mutateChatResponses();
}, [mutateChatResponses]);
return (
<Box
gap="1"
height="full"
minH="0"
display="flex"
flexDirec... | why do we need zIndex on this? Isn't it is the last layer by default? | }),
})) || []),
],
false
);
},
[mutateChatResponses]
);
const handleCreateChat = useCallback(() => {
mutateChatResponses();
}, [mutateChatResponses]);
const content = (
<>
{chats.map((chat) => (
<ChatListItem
key={chat.id}
... | @@ -79,6 +79,7 @@ export const ChatListBase = memo(function ChatListBase({
bg: "blackAlpha.400",
}}
{...props}
+ zIndex="var(--chakra-zIndices-base)" | website/src/components/Chat/ChatListBase.tsx | 26 | TypeScript | 0.214 | question | 69 | 51 | 51 | false | Bug Fix: Delete feature (Fixes three issues) | 3,383 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | CertifiedJoon | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatListBase.tsx
PR Title: Bug Fix: Delete feature (Fixes three issues)
BEFORE CODE: }),
})) || []),
],
false
);
},
[mutateChatResponses]
);
const handleCreateChat = useCa... | why do we need zIndex on this? Isn't it is the last layer by default? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
const { trigger: updateChatTitle, isMutating: isUpdatingTitle } = useSWRMutation(API_ROUTES.UPDATE_CHAT(), put);
const handleConfirmEdit = useCallback(async () => {
const title = inputRef.current?.value.trim();
if (!title) return;
await updateChatTitle({ chat_id: chat.id, title });
setIsEditing.off(... | Please revert this. Using a tag is much more better than JS | const handleConfirmEdit = useCallback(async () => {
const title = inputRef.current?.value.trim();
if (!title) return;
await updateChatTitle({ chat_id: chat.id, title });
setIsEditing.off();
onUpdateTitle({ chatId: chat.id, title });
}, [chat.id, onUpdateTitle, setIsEditing, updateChatTitle]);
... | @@ -78,7 +78,7 @@ export const ChatListItem = ({
<Button
// @ts-expect-error error due to dynamically changing as prop
ref={rootRef}
- {...(!isEditing ? { as: Link, href: ROUTES.CHAT(chat.id) } : { as: "div" })}
+ // {...(!isEditing ? { as: Link, href: ROUTES.CHAT(chat.id) } : { as: "div" }... | website/src/components/Chat/ChatListItem.tsx | 26 | TypeScript | 0.286 | suggestion | 59 | 51 | 51 | false | Bug Fix: Delete feature (Fixes three issues) | 3,383 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | CertifiedJoon | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatListItem.tsx
PR Title: Bug Fix: Delete feature (Fixes three issues)
BEFORE CODE: const { trigger: updateChatTitle, isMutating: isUpdatingTitle } = useSWRMutation(API_ROUTES.UPDATE_CHAT(), put);
const handleConfirmEdit = ... | Please revert this. Using a tag is much more better than JS | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import {
Button,
ButtonProps,
Modal,
ModalBody,
ModalCloseButton,
ModalContent,
ModalFooter,
ModalHeader,
ModalOverlay,
useDisclosure,
} from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
interface SkipButtonProps extends ButtonProps {
onSkip: () => void;
confirmSkip: boole... | Also can you only show the modal when the input is not empty? | import {
Button,
ButtonProps,
Modal,
ModalBody,
ModalCloseButton,
ModalContent,
ModalFooter,
ModalHeader,
ModalOverlay,
useDisclosure,
} from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
interface SkipButtonProps extends ButtonProps {
onSkip: () => void;
confirmSkip: boole... | @@ -0,0 +1,62 @@
+import {
+ Button,
+ ButtonProps,
+ Modal,
+ ModalBody,
+ ModalCloseButton,
+ ModalContent,
+ ModalFooter,
+ ModalHeader,
+ ModalOverlay,
+ useDisclosure,
+} from "@chakra-ui/react";
+import { useTranslation } from "next-i18next";
+
+interface SkipButtonProps extends ButtonProps {
+ onSkip:... | website/src/components/Buttons/Skip.tsx | 26 | TypeScript | 0.214 | question | 61 | 51 | 51 | false | Added confirmation dialog when skipping after entering text | 3,479 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | MehrJonas | Review this pull request change.
Language: TypeScript
File: website/src/components/Buttons/Skip.tsx
PR Title: Added confirmation dialog when skipping after entering text
BEFORE CODE: import {
Button,
ButtonProps,
Modal,
ModalBody,
ModalCloseButton,
ModalContent,
ModalFooter,
ModalHeader,
ModalOver... | Also can you only show the modal when the input is not empty? | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
s.add(id);
result.set(authType, s);
return result;
}, new Map());
const authOptions: AuthOptions = {
// Ensure we can store user data in a database.
adapter: PrismaAdapter(prisma),
providers,
session: {
strategy: "jwt",
},
pages: {
signIn: "/auth/signin",
verifyRequest: "/auth/verify",
... | You can receive image from user property in the callback here | result.set(authType, s);
return result;
}, new Map());
const authOptions: AuthOptions = {
// Ensure we can store user data in a database.
adapter: PrismaAdapter(prisma),
providers,
session: {
strategy: "jwt",
},
pages: {
signIn: "/auth/signin",
verifyRequest: "/auth/verify",
// error: "... | @@ -127,60 +126,44 @@ const authOptions: AuthOptions = {
session.user.role = token.role;
session.user.isNew = token.isNew;
session.user.name = token.name;
+ session.user.image = token.picture; | website/src/pages/api/auth/[...nextauth].ts | 26 | TypeScript | 0.286 | suggestion | 61 | 51 | 51 | false | Update profile image on login | 3,491 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/pages/api/auth/[...nextauth].ts
PR Title: Update profile image on login
BEFORE CODE: s.add(id);
result.set(authType, s);
return result;
}, new Map());
const authOptions: AuthOptions = {
// Ensure we can store user data in a database.
... | You can receive image from user property in the callback here | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
if (showEncourageMessage) {
onEncourageMessageClose();
}
},
[chatId, messageId, onEncourageMessageClose, onVote, score, showEncourageMessage]
);
const handleThumbsUp = useCallback(() => {
handleVote("+1");
}, [handleVote]);
const handleThumbsDown = useCallback(() => {
handl... | This probably has no effect, you can remove this. | const newScore = getNewScore(emoji, score);
onVote({ newScore, chatId, messageId, oldScore: score });
if (showEncourageMessage) {
onEncourageMessageClose();
}
},
[chatId, messageId, onEncourageMessageClose, onVote, score, showEncourageMessage]
);
const handleThumbsUp = useCa... | @@ -92,7 +92,9 @@ export const ChatMessageEntry = memo(function ChatMessageEntry({
const isAssistant = message.role === "assistant";
const [isEditing, setIsEditing] = useBoolean(false);
+ const [isPlainText, setIsPlainText] = useState<boolean>(false);
const inputRef = useRef<HTMLTextAreaElement>(null);
+ c... | website/src/components/Chat/ChatMessageEntry.tsx | 26 | TypeScript | 0.143 | refactor | 49 | 51 | 51 | false | Add: option to view plaintext chat message. Issue/2508 | 3,496 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | CertifiedJoon | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatMessageEntry.tsx
PR Title: Add: option to view plaintext chat message. Issue/2508
BEFORE CODE: if (showEncourageMessage) {
onEncourageMessageClose();
}
},
[chatId, messageId, onEncourageMessageClo... | This probably has no effect, you can remove this. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
bg: "gray.800",
}}
autoFocus
></Textarea>
</Box>
)}
{!isEditing && (
<Flex justifyContent={pagingSlot ? "space-between" : "end"} mt="1">
{pagingSlot}
{isAssistant && (
<MessageInlineEmojiRow... | Can you adjust icon base on the `isPlainText` state? I think using these two icons will make more sense
https://tabler-icons.io/i/markdown-off
https://tabler-icons.io/i/markdown
Just copy the SVG, we don't need to install the lib | borderRadius="xl"
_dark={{
bg: "gray.800",
}}
autoFocus
></Textarea>
</Box>
)}
{!isEditing && (
<Flex justifyContent={pagingSlot ? "space-between" : "end"} mt="1">
{pagingSlot}
... | @@ -175,6 +177,12 @@ export const ChatMessageEntry = memo(function ChatMessageEntry({
)}
{state === "complete" && (
<>
+ <BaseMessageEmojiButton | website/src/components/Chat/ChatMessageEntry.tsx | 26 | TypeScript | 0.571 | suggestion | 235 | 51 | 51 | false | Add: option to view plaintext chat message. Issue/2508 | 3,496 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | CertifiedJoon | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatMessageEntry.tsx
PR Title: Add: option to view plaintext chat message. Issue/2508
BEFORE CODE: bg: "gray.800",
}}
autoFocus
></Textarea>
</Box>
)}
... | Can you adjust icon base on the `isPlainText` state? I think using these two icons will make more sense
https://tabler-icons.io/i/markdown-off
https://tabler-icons.io/i/markdown
Just copy the SVG, we don't need to install the lib | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
tokenizer=pythia_tokenizer,
padding=True,
system_add_length=False,
)
features = [
create_dataset_entry_qa(
mode="sft",
questions=["Dummy Question?"],
answers=["Dummy Answer."],
)
]
batch = d(features)
expected_decoded_input_... | From a swollen rump, to "too" :joy: | tokenizer=pythia_tokenizer,
padding=True,
system_add_length=False,
)
features = [
create_dataset_entry_qa(
mode="sft",
questions=["Dummy Question?"],
answers=["Dummy Answer."],
)
]
batch = d(features)
expected_decoded_input_... | @@ -53,7 +53,7 @@ def test_dataset_entry_no_context(pythia_tokenizer):
# check if targets are as expected
assert pythia_tokenizer.decode(batch.targets[0]) == expected_decoded_targets[0]
- # check if masking is correct. Note that we mask the system aswell
+ # check if masking is correct. Note that we m... | model/model_training/tests/test_dialogue_data_collator.py | 26 | Python | 0.071 | suggestion | 35 | 51 | 51 | false | fix typos | 3,543 | LAION-AI/Open-Assistant | 29,137 | Python | bitplane | RainRat | Review this pull request change.
Language: Python
File: model/model_training/tests/test_dialogue_data_collator.py
PR Title: fix typos
BEFORE CODE: tokenizer=pythia_tokenizer,
padding=True,
system_add_length=False,
)
features = [
create_dataset_entry_qa(
mode="sft"... | From a swollen rump, to "too" :joy: | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import torch
# rotary pos emb helpers (torch.jit.script does not seem to support staticmethod...)
def rotate_half(x):
x1, x2 = x[..., : x.shape[-1] // 2], x[..., x.shape[-1] // 2 :]
return torch.cat((-x2, x1), dim=x1.ndim - 1) # dim=-1 triggers a bug in torch < 1.8.0
class RWNTKScaledRope(torch.nn.Module):... | Could you add some comments about the origin of these rope scaling variants? Were they copied from the [scaled-rope](https://github.com/jquesnelle/scaled-rope) repository? If they were imported from a different place we should add a proper attribution. | import torch
# rotary pos emb helpers (torch.jit.script does not seem to support staticmethod...)
def rotate_half(x):
x1, x2 = x[..., : x.shape[-1] // 2], x[..., x.shape[-1] // 2 :]
return torch.cat((-x2, x1), dim=x1.ndim - 1) # dim=-1 triggers a bug in torch < 1.8.0
class RWNTKScaledRope(torch.nn.Module):... | @@ -0,0 +1,175 @@
+import torch
+
+
+# rotary pos emb helpers (torch.jit.script does not seem to support staticmethod...)
+def rotate_half(x):
+ x1, x2 = x[..., : x.shape[-1] // 2], x[..., x.shape[-1] // 2 :]
+ return torch.cat((-x2, x1), dim=x1.ndim - 1) # dim=-1 triggers a bug in torch < 1.8.0
+
+
+class RWNTK... | model/model_training/models/rope.py | 10 | Python | 0.5 | question | 252 | 35 | 35 | false | Add RoPE Interpolation | 3,564 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | shahules786 | Review this pull request change.
Language: Python
File: model/model_training/models/rope.py
PR Title: Add RoPE Interpolation
BEFORE CODE: import torch
# rotary pos emb helpers (torch.jit.script does not seem to support staticmethod...)
def rotate_half(x):
x1, x2 = x[..., : x.shape[-1] // 2], x[..., x.shape[-1... | Could you add some comments about the origin of these rope scaling variants? Were they copied from the [scaled-rope](https://github.com/jquesnelle/scaled-rope) repository? If they were imported from a different place we should add a proper attribution. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
if training_conf.use_custom_sampler:
samples_length = None
if training_conf.sort_by_length:
samples_length = list(
map(
lambda x: train_collate_fn.process_one(x, return_length=True),
tqdm(train, desc="Calculating lengths per sample... | Is there a benefit of late import, otherwise I would recommend to move this to the other imports at the beginning of the file (to make it easier to see all main dependencies). |
if training_conf.use_custom_sampler:
samples_length = None
if training_conf.sort_by_length:
samples_length = list(
map(
lambda x: train_collate_fn.process_one(x, return_length=True),
tqdm(train, desc="Calculating lengths per sample... | @@ -416,9 +416,14 @@ def main():
sampler = None
metrics, preprocess_fns = get_metrics(training_conf, tokenizer)
-
model = get_model(training_conf, tokenizer)
+ from model_training.models.patching import RopePatch | model/model_training/trainer_sft.py | 26 | Python | 0.429 | refactor | 175 | 51 | 51 | false | Add RoPE Interpolation | 3,564 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | shahules786 | Review this pull request change.
Language: Python
File: model/model_training/trainer_sft.py
PR Title: Add RoPE Interpolation
BEFORE CODE:
if training_conf.use_custom_sampler:
samples_length = None
if training_conf.sort_by_length:
samples_length = list(
map(
... | Is there a benefit of late import, otherwise I would recommend to move this to the other imports at the beginning of the file (to make it easier to see all main dependencies). | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
@property
def num_requests_in_progress(self) -> int:
return len(self.work_request_map)
@classmethod
async def build(cls, websocket: fastapi.WebSocket, api_key: str, protocol_version: str):
try:
worker_utils.get_protocol_version(protocol_version)
api_key = worker... | Moved to `HandleWorkerContext.build()` without changes | @property
def num_requests_in_progress(self) -> int:
return len(self.work_request_map)
@classmethod
async def build(cls, websocket: fastapi.WebSocket, api_key: str, protocol_version: str):
try:
worker_utils.get_protocol_version(protocol_version)
api_key = worker_... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.571 | suggestion | 54 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE:
@property
def num_requests_in_progress(self) -> int:
return len(self.work_request_map)
... | Moved to `HandleWorkerContext.build()` without changes | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
@classmethod
async def build(cls, websocket: fastapi.WebSocket, api_key: str, protocol_version: str):
try:
worker_utils.get_protocol_version(protocol_version)
api_key = worker_utils.get_api_key(api_key)
worker_id = await worker_utils.get_worker_id(api_key=api_key, pro... | Moved to `SessionManager.init()` | async def build(cls, websocket: fastapi.WebSocket, api_key: str, protocol_version: str):
try:
worker_utils.get_protocol_version(protocol_version)
api_key = worker_utils.get_api_key(api_key)
worker_id = await worker_utils.get_worker_id(api_key=api_key, protocol_version=pro... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.286 | suggestion | 32 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE: @classmethod
async def build(cls, websocket: fastapi.WebSocket, api_key: str, protocol_version: ... | Moved to `SessionManager.init()` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
except fastapi.HTTPException as e:
logger.warning(f"handle_worker: {e.status_code=} {e.detail=}")
if e.status_code == fastapi.status.HTTP_426_UPGRADE_REQUIRED:
await worker_utils.send_worker_request(websocket=websocket, request=inference.UpgradeProtocolRequest())
... | Moved to `SessionManager.update()` | logger.warning(f"handle_worker: {e.status_code=} {e.detail=}")
if e.status_code == fastapi.status.HTTP_426_UPGRADE_REQUIRED:
await worker_utils.send_worker_request(websocket=websocket, request=inference.UpgradeProtocolRequest())
elif e.status_code == fastapi.status.HT... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.286 | suggestion | 34 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE: except fastapi.HTTPException as e:
logger.warning(f"handle_worker: {e.status_code=} ... | Moved to `SessionManager.update()` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
await worker_utils.send_worker_request(websocket=websocket, request=inference.WrongApiKeyRequest())
try:
await websocket.close(code=e.status_code, reason=e.detail)
except Exception:
pass
raise fastapi.WebSocketException(e.status_code, e... | Moved to `FuturesManager.ensure_listening_for_work_requests()`. The future is now wrapped such that it returns a `tuple[FutureType, X]` where `X` is the return type of whatever the future returned before. | try:
await websocket.close(code=e.status_code, reason=e.detail)
except Exception:
pass
raise fastapi.WebSocketException(e.status_code, e.detail)
logger.info(f"handle_worker: {worker_id=}")
worker_info = await worker_utils.receive_worke... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.786 | suggestion | 204 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE: await worker_utils.send_worker_request(websocket=websocket, request=inference.WrongApiKe... | Moved to `FuturesManager.ensure_listening_for_work_requests()`. The future is now wrapped such that it returns a `tuple[FutureType, X]` where `X` is the return type of whatever the future returned before. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
except Exception:
pass
raise fastapi.WebSocketException(e.status_code, e.detail)
logger.info(f"handle_worker: {worker_id=}")
worker_info = await worker_utils.receive_worker_info(websocket)
logger.info(f"handle_worker: {worker_info=}")
work_queue... | Moved to `FuturesManager.ensure_listening_for_worker_response()` with similar changes to return the `FutureType` in addition to the response. | pass
raise fastapi.WebSocketException(e.status_code, e.detail)
logger.info(f"handle_worker: {worker_id=}")
worker_info = await worker_utils.receive_worker_info(websocket)
logger.info(f"handle_worker: {worker_info=}")
work_queue = queueing.work_queue(deps.red... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.714 | suggestion | 141 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE: except Exception:
pass
raise fastapi.WebSocketException(e.status... | Moved to `FuturesManager.ensure_listening_for_worker_response()` with similar changes to return the `FutureType` in addition to the response. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
return cls(
websocket=websocket,
worker_id=worker_id,
worker_info=worker_info,
work_queue=work_queue,
blocking_work_queue=blocking_work_queue
)
async def close(self):
try:
await self.blocking_work_queue.redis_client.cl... | Moved to `FuturesManager.wait_for_event_or_timeout()` | websocket=websocket,
worker_id=worker_id,
worker_info=worker_info,
work_queue=work_queue,
blocking_work_queue=blocking_work_queue,
)
async def close(self):
try:
await self.blocking_work_queue.redis_client.close()
except... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.429 | suggestion | 53 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE:
return cls(
websocket=websocket,
worker_id=worker_id,
worke... | Moved to `FuturesManager.wait_for_event_or_timeout()` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
worker_info=worker_info,
work_queue=work_queue,
blocking_work_queue=blocking_work_queue
)
async def close(self):
try:
await self.blocking_work_queue.redis_client.close()
except Exception:
logger.warning("Error while closing redis c... | This is the test that has so far been used to identify a dequeued work request, which has now been replaced by a check for `type_ == FutureType.WORK_REQUEST` | blocking_work_queue=blocking_work_queue,
)
async def close(self):
try:
await self.blocking_work_queue.redis_client.close()
except Exception:
logger.warning("Error while closing redis client")
try:
await self.websocket.close()
... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.714 | suggestion | 157 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE: worker_info=worker_info,
work_queue=work_queue,
blocking_work_queue=... | This is the test that has so far been used to identify a dequeued work request, which has now been replaced by a check for `type_ == FutureType.WORK_REQUEST` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
WORK_REQUEST = 'WORK_REQUEST'
WORKER_RESPONSE = 'WORKER_RESPONSE'
AwaitResult = TypeVar('AwaitResult')
class FuturesManager:
_context: HandleWorkerContext
_futures: set[asyncio.Future]
def __init__(self, context: HandleWorkerContext):
self._context = context
self._futures = set()... | Moved to `_handle_work_request` |
AwaitResult = TypeVar("AwaitResult")
class FuturesManager:
_context: HandleWorkerContext
_futures: set[asyncio.Future]
def __init__(self, context: HandleWorkerContext):
self._context = context
self._futures = set()
def ensure_listening_for_work_requests(self):
requests_in_pr... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.429 | suggestion | 31 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE: WORK_REQUEST = 'WORK_REQUEST'
WORKER_RESPONSE = 'WORKER_RESPONSE'
AwaitResult = TypeVar('AwaitR... | Moved to `_handle_work_request` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
for ftr in self._futures:
try:
ftr.cancel()
except Exception:
logger.warning("Error while cancelling pending future")
def _as_typed_future(self, future_type: FutureType, awaitable: Awaitable[AwaitResult]) -> asyncio.Future:
async def wrapp... | Moved to `_handle_worker_response()` | logger.info(f"Cancelling {len(self._futures)} pending futures")
for ftr in self._futures:
try:
ftr.cancel()
except Exception:
logger.warning("Error while cancelling pending future")
def _as_typed_future(self, future_type: FutureType, awaitable... | @@ -85,191 +72,319 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.429 | suggestion | 36 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE: for ftr in self._futures:
try:
ftr.cancel()
except Excep... | Moved to `_handle_worker_response()` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
async def initiate_work_for_message(
*,
websocket: fastapi.WebSocket,
work_queue: queueing.RedisQueue,
message_id: str,
worker_id: str,
worker_config: inference.WorkerConfig,
) -> inference.WorkRequest:
async with deps.manual_create_session() as session:
cr = chat_repository.ChatRep... | In order to avoid logical changes as much as possible, I kept this piece of code here as it was, but that now means that it accesses the internal structure of the `FuturesManager`.
I'm considering whether it might be nicer to not throw all the futures into a single set, but instead keep one future in a specific vari... | await session.update(worker_response.metrics)
case "general_error":
worker_response = cast(inference.GeneralErrorResponse, worker_response)
await handle_general_error_response(
response=worker_response,
)
awa... | @@ -85,191 +72,321 @@ def get_work_request_container(work_request_map: WorkRequestContainerMap, reques
return container
-@router.websocket("/work")
-async def handle_worker(
- websocket: fastapi.WebSocket,
- api_key: str = worker_utils.api_key_header,
- protocol_version: str = worker_utils.protocol_ve... | inference/server/oasst_inference_server/routes/workers.py | 26 | Python | 0.786 | suggestion | 655 | 51 | 51 | false | Cleanup `handle_worker()` in preparation for #2815 (Stop generation early) | 3,573 | LAION-AI/Open-Assistant | 29,137 | Python | 0xfacade | 0xfacade | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Cleanup `handle_worker()` in preparation for #2815 (Stop generation early)
BEFORE CODE:
async def initiate_work_for_message(
*,
websocket: fastapi.WebSocket,
work_queue: queueing.... | In order to avoid logical changes as much as possible, I kept this piece of code here as it was, but that now means that it accesses the internal structure of the `FuturesManager`.
I'm considering whether it might be nicer to not throw all the futures into a single set, but instead keep one future in a specific vari... | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import transformers
# from .gptj import get_model as get_gptj_model
SUPPORTED_MODELS = ["galactica", "gpt-j"]
def freeze_top_n_layers(model, target_layers):
# its possible we can simply detect which module is a ModuleList
# and simply freeze the module without doing string parsing
for name, param in mod... | so falcon 40B still needs trust_remote_code = True. |
def get_specific_model(
model_name,
seq2seqmodel=False,
without_head=False,
cache_dir=".cache",
quantization=False,
**kwargs,
):
if without_head:
model = transformers.AutoModel.from_pretrained(model_name, cache_dir=cache_dir, **kwargs)
elif seq2seqmodel:
# encoder-decod... | @@ -25,17 +21,27 @@ def freeze_top_n_layers(model, target_layers):
def get_specific_model(
- model_name, seq2seqmodel=False, without_head=False, cache_dir=".cache", quantization=False, **kwargs
+ model_name,
+ seq2seqmodel=False,
+ without_head=False,
+ cache_dir=".cache",
+ quantization=False,
+... | model/model_training/models/__init__.py | 26 | Python | 0.214 | suggestion | 51 | 42 | 28 | false | Add flash-attention patch for falcon-7b | 3,580 | LAION-AI/Open-Assistant | 29,137 | Python | jordiclive | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/models/__init__.py
PR Title: Add flash-attention patch for falcon-7b
BEFORE CODE: import transformers
# from .gptj import get_model as get_gptj_model
SUPPORTED_MODELS = ["galactica", "gpt-j"]
def freeze_top_n_layers(model, target_layer... | so falcon 40B still needs trust_remote_code = True. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import torch
import torch.nn.functional as F
def compute_flash_attention(flash_attn, q, k, v, attention_mask=None, head_mask=None):
# q, k, v: [bs, seq_len, num_attention_heads, attn_head_size]
# attention_mask (float): [bs, seq_len]
batch_size, max_len = q.size(0), q.size(1)
qkv = torch.stack([q, k,... | Falcon is torch.bfloat16? Will the casting to float16 pose a problem. | import torch
import torch.nn.functional as F
def compute_flash_attention(flash_attn, q, k, v, attention_mask=None, head_mask=None):
# q, k, v: [bs, seq_len, num_attention_heads, attn_head_size]
# attention_mask (float): [bs, seq_len]
batch_size, max_len = q.size(0), q.size(1)
qkv = torch.stack([q, k,... | @@ -7,11 +7,14 @@ def compute_flash_attention(flash_attn, q, k, v, attention_mask=None, head_mask=
# attention_mask (float): [bs, seq_len]
batch_size, max_len = q.size(0), q.size(1)
- qkv = torch.stack([q, k, v], dim=2).to(torch.float16) # need to truncate in case input is fp32
+ qkv = torch.stack([q... | model/model_training/models/patching_utils.py | 13 | Python | 0.357 | suggestion | 69 | 38 | 38 | false | Add flash-attention patch for falcon-7b | 3,580 | LAION-AI/Open-Assistant | 29,137 | Python | jordiclive | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/models/patching_utils.py
PR Title: Add flash-attention patch for falcon-7b
BEFORE CODE: import torch
import torch.nn.functional as F
def compute_flash_attention(flash_attn, q, k, v, attention_mask=None, head_mask=None):
# q, k, v: [b... | Falcon is torch.bfloat16? Will the casting to float16 pose a problem. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
def make_prompt_and_parameters(
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
) -> tuple[str, interface.GenerateStreamParameters]:
"""Prepare a formatted prompt and stream generation parameters based on a work request."""
if settings.oa_protocol_version != "v2":
... | Can we change this so that the whole custom instruction prefix is not prepended at all if no instructions are given? Otherwise we are just using up tokens for no reason, as the worker retains the system prompt in context always at the cost of other tokens |
def make_prompt_and_parameters(
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
) -> tuple[str, interface.GenerateStreamParameters]:
"""Prepare a formatted prompt and stream generation parameters based on a work request."""
if settings.oa_protocol_version != "v2":
... | @@ -40,9 +41,17 @@ def _prepare_message(message: inference.MessageRead) -> str:
# Construct prompt
messages = [_prepare_message(message) for message in work_request.thread.messages]
- # Prepend system prompt if it was specified in work parameters
- if work_request.parameters.system_prompt:
- pr... | inference/worker/work.py | 26 | Python | 0.5 | question | 255 | 51 | 51 | false | Custom instructions feature | 3,597 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | draganjovanovich | Review this pull request change.
Language: Python
File: inference/worker/work.py
PR Title: Custom instructions feature
BEFORE CODE:
def make_prompt_and_parameters(
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
) -> tuple[str, interface.GenerateStreamParameters]:
"""P... | Can we change this so that the whole custom instruction prefix is not prepended at all if no instructions are given? Otherwise we are just using up tokens for no reason, as the worker retains the system prompt in context always at the cost of other tokens | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
try:
original_prompt = work_request.thread.messages[-1].content
if not original_prompt:
raise ValueError("Prompt is empty")
language = "English"
plugin = next((p for p in parameters.plugins if p.enabled), None)
tools_instructions_template, tools = compose_tools_... | Also here - surely if `user_profile` and `user_response_instructions` are empty, we want this value `custom_instructions` to also be an entirely empty string, without the contents of `CUSTOM_INSTRUCTIONS_PREFIX` | try:
original_prompt = work_request.thread.messages[-1].content
if not original_prompt:
raise ValueError("Prompt is empty")
language = "English"
plugin = next((p for p in parameters.plugins if p.enabled), None)
tools_instructions_template, tools = compose_tools_... | @@ -355,11 +379,17 @@ def handle_conversation(
"language",
"current_time",
"action_input_format",
+ "custom_instructions",
] + (["tools_names"] if plugin_enabled else [])
# TODO: Consider passing language from the UI here
prompt_template = P... | inference/worker/chat_chain.py | 26 | Python | 0.643 | suggestion | 211 | 51 | 51 | false | Custom instructions feature | 3,597 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | draganjovanovich | Review this pull request change.
Language: Python
File: inference/worker/chat_chain.py
PR Title: Custom instructions feature
BEFORE CODE: try:
original_prompt = work_request.thread.messages[-1].content
if not original_prompt:
raise ValueError("Prompt is empty")
language = "E... | Also here - surely if `user_profile` and `user_response_instructions` are empty, we want this value `custom_instructions` to also be an entirely empty string, without the contents of `CUSTOM_INSTRUCTIONS_PREFIX` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
def make_prompt_and_parameters(
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
) -> tuple[str, interface.GenerateStreamParameters]:
"""Prepare a formatted prompt and stream generation parameters based on a work request."""
if settings.oa_protocol_version != "v2":
... | This can be simplified a bit (currently the same `if` condition is repeated)
```suggestion
work_params = work_request.parameters
if work_params.system_prompt or work_params.user_profile or work_params.user_response_instructions:
pre_prompt = V2_SYSTEM_PREFIX + (work_params.system_prompt or "")
``` | def make_prompt_and_parameters(
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
) -> tuple[str, interface.GenerateStreamParameters]:
"""Prepare a formatted prompt and stream generation parameters based on a work request."""
if settings.oa_protocol_version != "v2":
... | @@ -40,9 +41,20 @@ def _prepare_message(message: inference.MessageRead) -> str:
# Construct prompt
messages = [_prepare_message(message) for message in work_request.thread.messages]
- # Prepend system prompt if it was specified in work parameters
- if work_request.parameters.system_prompt:
- pr... | inference/worker/work.py | 26 | Python | 0.857 | suggestion | 319 | 51 | 51 | false | Custom instructions feature | 3,597 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | draganjovanovich | Review this pull request change.
Language: Python
File: inference/worker/work.py
PR Title: Custom instructions feature
BEFORE CODE: def make_prompt_and_parameters(
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
) -> tuple[str, interface.GenerateStreamParameters]:
"""Pr... | This can be simplified a bit (currently the same `if` condition is repeated)
```suggestion
work_params = work_request.parameters
if work_params.system_prompt or work_params.user_profile or work_params.user_response_instructions:
pre_prompt = V2_SYSTEM_PREFIX + (work_params.system_prompt or "")
``` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
const resetParameters = (setValue: UseFormSetValue<ChatConfigFormData>, params: SamplingParameters) => {
for (const [key, value] of Object.entries(params) as Array<[keyof SamplingParameters, number]>) {
setValue(key, value); // call setValue instead of setValues to avoid reset unwanted fields
}
};
export cons... | This state can be handled by react-hook-form. You just need to wrap the `CustomInstructions` with the `Controller` component. | };
const findPresetName = (presets: ModelParameterConfig[], config: SamplingParameters) => {
return (
presets.find((preset) => areParametersEqual(preset.sampling_parameters, config))?.name ?? unKnownCustomPresetName
);
};
const resetParameters = (setValue: UseFormSetValue<ChatConfigFormData>, params: Sampling... | @@ -104,6 +111,10 @@ export const ChatConfigForm = memo(function ChatConfigForm() {
const selectedModel = getValues("model_config_name"); // have to use getValues to here to access latest value
const selectedPlugins = getValues("plugins");
const presets = modelInfos.find((model) => model.name === selectedModel... | website/src/components/Chat/ChatConfigForm.tsx | 26 | TypeScript | 0.643 | suggestion | 125 | 51 | 51 | false | Custom instructions feature | 3,597 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | draganjovanovich | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConfigForm.tsx
PR Title: Custom instructions feature
BEFORE CODE:
const resetParameters = (setValue: UseFormSetValue<ChatConfigFormData>, params: SamplingParameters) => {
for (const [key, value] of Object.entries(params) ... | This state can be handled by react-hook-form. You just need to wrap the `CustomInstructions` with the `Controller` component. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import gzip
import json
import re
from pathlib import Path
from typing import List, Optional, Union
import requests
from datasets import load_dataset
from model_training.custom_datasets.formatting import DatasetEntrySft, Role, Utterance
from model_training.custom_datasets.oasst_dataset import ListDataset
from model_tr... | I should have used shuffle(seed) too in my code ;-) .. will update later. |
def __getitem__(self, idx):
conversation, instruction = [self.dataset[idx][key] for key in ("conversation", "instruction")]
conversation = [(item["input"], item["output"]) for item in conversation]
conversation = list(sum(conversation, ()))
conv_utt: list[Utterance] = [
... | @@ -193,3 +194,61 @@ def __getitem__(self, idx):
]
return DatasetEntrySft(conversation=conv_utt, system_message=instruction)
+
+
+class DolphinMix(Dataset):
+ name = "dophin-mix"
+
+ def __init__(self, cache_dir, num_samples=100000, max_char_len=8000, seed=42):
+ self.dataset = load_dat... | model/model_training/custom_datasets/prompt_dialogue.py | 26 | Python | 0.286 | suggestion | 73 | 196 | 51 | false | Dolphin gpt 3.5 data mix | 3,606 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | shahules786 | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/prompt_dialogue.py
PR Title: Dolphin gpt 3.5 data mix
BEFORE CODE: import gzip
import json
import re
from pathlib import Path
from typing import List, Optional, Union
import requests
from datasets import load_dataset
from ... | I should have used shuffle(seed) too in my code ;-) .. will update later. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
import gzip
import json
import re
from pathlib import Path
from typing import List, Optional, Union
import requests
from datasets import load_dataset
from model_training.custom_datasets.formatting import DatasetEntrySft, Role, Utterance
from model_training.custom_datasets.oasst_dataset import ListDataset
from model_tr... | Generating the indices to pick set every time seems a bit costly .. but since it is only done during startup it might be ok. | name = "dophin-mix"
def __init__(self, cache_dir, num_samples=100000, max_char_len=8000, seed=42):
self.dataset = load_dataset(
"ehartford/dolphin", data_files="flan5m-alpaca-uncensored.jsonl", cache_dir=cache_dir
)
self.dataset = self.dataset["train"].shuffle(seed).select(r... | @@ -193,3 +194,61 @@ def __getitem__(self, idx):
]
return DatasetEntrySft(conversation=conv_utt, system_message=instruction)
+
+
+class DolphinMix(Dataset):
+ name = "dophin-mix"
+
+ def __init__(self, cache_dir, num_samples=100000, max_char_len=8000, seed=42):
+ self.dataset = load_dat... | model/model_training/custom_datasets/prompt_dialogue.py | 26 | Python | 0.357 | suggestion | 125 | 196 | 51 | false | Dolphin gpt 3.5 data mix | 3,606 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | shahules786 | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/prompt_dialogue.py
PR Title: Dolphin gpt 3.5 data mix
BEFORE CODE: import gzip
import json
import re
from pathlib import Path
from typing import List, Optional, Union
import requests
from datasets import load_dataset
from ... | Generating the indices to pick set every time seems a bit costly .. but since it is only done during startup it might be ok. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
self.dataset = load_dataset("shahules786/orca-chat", split="train", data_files=data_files, cache_dir=cache_dir)
def __len__(self):
return len(self.dataset)
def __getitem__(self, idx):
conversation, instruction = [self.dataset[idx][key] for key in ("conversation", "instruction")]
... | Isn't it better to add `data_files` as an argument with the default value "flan-5m..."? In this way, we can change to gpt-4 version using config if required.
` | self.dataset = load_dataset("shahules786/orca-chat", split="train", data_files=data_files, cache_dir=cache_dir)
def __len__(self):
return len(self.dataset)
def __getitem__(self, idx):
conversation, instruction = [self.dataset[idx][key] for key in ("conversation", "instruction")]
... | @@ -199,45 +198,37 @@ def __getitem__(self, idx):
class DolphinMix(Dataset):
name = "dophin-mix"
- def __init__(self, cache_dir, num_samples=100000, max_char_len=8000, seed=42):
+ def __init__(self, cache_dir, num_samples: Optional[int] = None, max_char_len: int = 8000, seed: int = 42): | model/model_training/custom_datasets/prompt_dialogue.py | 26 | Python | 0.571 | question | 162 | 51 | 51 | false | Changes for orcacode experiment | 3,612 | LAION-AI/Open-Assistant | 29,137 | Python | shahules786 | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/prompt_dialogue.py
PR Title: Changes for orcacode experiment
BEFORE CODE: self.dataset = load_dataset("shahules786/orca-chat", split="train", data_files=data_files, cache_dir=cache_dir)
def __len__(self):
... | Isn't it better to add `data_files` as an argument with the default value "flan-5m..."? In this way, we can change to gpt-4 version using config if required.
` | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
</Box>
</MenuItem>
</MenuList>
</Menu>
<Modal isOpen={isOpen} onClose={onClose} size="xl" initialFocusRef={textareaRef}>
<ModalOverlay />
<ModalContent>
<ModalHeader>{t(isCreatePlugin ? "add_plugin" : "edit_plugin")}</ModalHeader>
<ModalClose... | if you have a url for installing plugins you could use getitAI's callback mechanism: https://www.getit.ai/gpt-plugins/chat-integration | }}
/>
</Box>
</MenuItem>
</MenuList>
</Menu>
<Modal isOpen={isOpen} onClose={onClose} size="xl" initialFocusRef={textareaRef}>
<ModalOverlay />
<ModalContent>
<ModalHeader>{t(isCreatePlugin ? "add_plugin" : "edit_plugin")}... | @@ -234,6 +236,16 @@ export const PluginsChooser = ({ plugins, setPlugins }: PluginsChooserProps) =>
placeholder={t("plugin_url_placeholder")}
/>
{selectedForEditPluginIndex !== null && <JsonCard>{plugins[selectedForEditPluginIndex]}</JsonCard>}
+ <Box>
+ ... | website/src/components/Chat/PluginsChooser.tsx | 26 | TypeScript | 0.429 | suggestion | 134 | 51 | 51 | false | Added plugin repositories to add-plugin menu | 3,638 | LAION-AI/Open-Assistant | 29,137 | Python | digi604 | yk | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/PluginsChooser.tsx
PR Title: Added plugin repositories to add-plugin menu
BEFORE CODE: </Box>
</MenuItem>
</MenuList>
</Menu>
<Modal isOpen={isOpen} onClose={onClose} size="xl" initialFo... | if you have a url for installing plugins you could use getitAI's callback mechanism: https://www.getit.ai/gpt-plugins/chat-integration | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
case "REVIEW":
return { mode: "EDIT", replyValidity: "VALID" };
case "DEFAULT_WARN":
return { mode: "EDIT", replyValidity: "DEFAULT" };
case "SUBMITTED":
// allow return to edit from subbmitted mode (error happen during submitting task)
... | No issues found. | case "REVIEW":
return { mode: "EDIT", replyValidity: "VALID" };
case "DEFAULT_WARN":
return { mode: "EDIT", replyValidity: "DEFAULT" };
case "SUBMITTED":
// allow return to edit from subbmitted mode (error happen during submitting task)
... | @@ -1,7 +1,6 @@
import { useToast } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
-import { useCallback, useEffect, useReducer } from "react";
-import { useMemo, useRef } from "react";
+import { useCallback, useEffect, useMemo, useReducer, useRef } from "react";
import { TaskControls } from... | website/src/components/Tasks/Task/Task.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Implemented ability to press ctrl+enter (or cmd+enter) to Review/Subm… | 1,237 | LAION-AI/Open-Assistant | 29,137 | Python | cooperaustinj | Review this pull request change.
Language: TypeScript
File: website/src/components/Tasks/Task/Task.tsx
PR Title: Implemented ability to press ctrl+enter (or cmd+enter) to Review/Subm…
BEFORE CODE: case "REVIEW":
return { mode: "EDIT", replyValidity: "VALID" };
case "DEFAULT_WAR... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import {
Button,
Flex,
Modal,
ModalBody,
ModalCloseButton,
ModalContent,
ModalFooter,
ModalHeader,
ModalOverlay,
} from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
interface UnchangedWarningProps {
show: boolean;
title: string;
message: string;
continueButtonText: strin... | No issues found. | import {
Button,
Flex,
Modal,
ModalBody,
ModalCloseButton,
ModalContent,
ModalFooter,
ModalHeader,
ModalOverlay,
} from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
interface UnchangedWarningProps {
show: boolean;
title: string;
message: string;
continueButtonText: strin... | @@ -9,6 +9,7 @@ import {
ModalHeader,
ModalOverlay,
} from "@chakra-ui/react";
+import { useTranslation } from "next-i18next";
interface UnchangedWarningProps {
show: boolean;
@@ -20,6 +21,8 @@ interface UnchangedWarningProps {
}
export const UnchangedWarning = (props: UnchangedWarningProps) => {
+ co... | website/src/components/Tasks/UnchangedWarning.tsx | 0 | TypeScript | 1 | none | 16 | 47 | 47 | true | Localize some labels that were missing in the tasks view | 1,607 | LAION-AI/Open-Assistant | 29,137 | Python | GuilleHoardings | Review this pull request change.
Language: TypeScript
File: website/src/components/Tasks/UnchangedWarning.tsx
PR Title: Localize some labels that were missing in the tasks view
BEFORE CODE: import {
Button,
Flex,
Modal,
ModalBody,
ModalCloseButton,
ModalContent,
ModalFooter,
ModalHeader,
ModalOver... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
except OasstError:
raise
except Exception:
logger.exception("Failed to generate task..")
raise OasstError("Failed to generate task.", OasstErrorCode.TASK_GENERATION_FAILED)
return task
@router.post("/availability", response_model=dict[protocol_schema.TaskRequestType, int])
def tas... | No issues found. |
except OasstError:
raise
except Exception:
logger.exception("Failed to generate task..")
raise OasstError("Failed to generate task.", OasstErrorCode.TASK_GENERATION_FAILED)
return task
@router.post("/availability", response_model=dict[protocol_schema.TaskRequestType, int])
def tas... | @@ -34,6 +34,24 @@
minutes=settings.RATE_LIMIT_TASK_API_MINUTES,
)
),
+ Depends(
+ deps.UserTaskTypeRateLimiter(
+ [
+ protocol_schema.TaskRequestType.assistant_reply,
+ ],
+ times=settings.RATE_LIMI... | backend/oasst_backend/api/v1/tasks.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Implement separate rate limit for message submission tasks | 1,747 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: backend/oasst_backend/api/v1/tasks.py
PR Title: Implement separate rate limit for message submission tasks
BEFORE CODE:
except OasstError:
raise
except Exception:
logger.exception("Failed to generate task..")
raise OasstError("... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
hugging_face_api.headers: Dict[str, str] = {"Authorization": f"Bearer {hugging_face_api.api_key}"}
result = await hugging_face_api.post(text)
return result
@app.task(name="toxicity")
def toxicity(text, message_id, api_client):
try:
logger.info(f"checking toxicity : {api_client}")
with... | No issues found. | hugging_face_api.headers: Dict[str, str] = {"Authorization": f"Bearer {hugging_face_api.api_key}"}
result = await hugging_face_api.post(text)
return result
@app.task(name="toxicity")
def toxicity(text, message_id, api_client):
try:
logger.info(f"checking toxicity : {api_client}")
with... | @@ -0,0 +1,135 @@
+from __future__ import absolute_import, unicode_literals
+
+import os
+from datetime import datetime
+from typing import Any, Dict, List
+
+from asgiref.sync import async_to_sync
+from celery import Celery, shared_task
+from dotenv import load_dotenv
+from loguru import logger
+from oasst_backend.mod... | backend/oasst_backend/celery_worker/scheduled_tasks.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | feature: Using Celery for APIs and periodic tasks | 1,754 | LAION-AI/Open-Assistant | 29,137 | Python | melvinebenezer | Review this pull request change.
Language: Python
File: backend/oasst_backend/celery_worker/scheduled_tasks.py
PR Title: feature: Using Celery for APIs and periodic tasks
BEFORE CODE: hugging_face_api.headers: Dict[str, str] = {"Authorization": f"Bearer {hugging_face_api.api_key}"}
result = await hugging_fa... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import Image from "next/image";
export function UserAvatar(options: { displayName: string; avatarUrl: string | null }) {
const { displayName, avatarUrl } = options;
return (
<>
<Image
src={
avatarUrl
? avatarUrl
: `https://api.dicebear.com/5.x/initials/png?seed=${... | No issues found. | import Image from "next/image";
export function UserAvatar(options: { displayName: string; avatarUrl: string | null }) {
const { displayName, avatarUrl } = options;
return (
<>
<Image
src={
avatarUrl
? avatarUrl
: `https://api.dicebear.com/5.x/initials/png?seed=${... | @@ -0,0 +1,18 @@
+import Image from "next/image";
+export function UserAvatar(options: { displayName: string; avatarUrl: string | null }) {
+ const { displayName, avatarUrl } = options;
+ return (
+ <>
+ <Image
+ src={
+ avatarUrl
+ ? avatarUrl
+ : `https://api.dicebear.c... | website/src/components/UserAvatar.tsx | 0 | TypeScript | 1 | none | 16 | 19 | 19 | true | Display user avatar in leaderboard | 1,767 | LAION-AI/Open-Assistant | 29,137 | Python | mrloldev | Review this pull request change.
Language: TypeScript
File: website/src/components/UserAvatar.tsx
PR Title: Display user avatar in leaderboard
BEFORE CODE: import Image from "next/image";
export function UserAvatar(options: { displayName: string; avatarUrl: string | null }) {
const { displayName, avatarUrl } = o... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { Box, Button, Center, Link, Text, theme, useColorModeValue } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { AlertTriangle } from "lucide-react";
import Head from "next/head";
import { EmptyState } from "src/components/EmptyState";
import { getTransparentHeaderLayout } from "src... | No issues found. | import { Box, Button, Center, Link, Text, theme, useColorModeValue } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { AlertTriangle } from "lucide-react";
import Head from "next/head";
import { EmptyState } from "src/components/EmptyState";
import { getTransparentHeaderLayout } from "src... | @@ -1,11 +1,13 @@
import { Box, Button, Center, Link, Text, theme, useColorModeValue } from "@chakra-ui/react";
+import { useTranslation } from "next-i18next";
import { AlertTriangle } from "lucide-react";
import Head from "next/head";
import { EmptyState } from "src/components/EmptyState";
import { getTransparent... | website/src/pages/404.tsx | 0 | TypeScript | 1 | none | 16 | 48 | 48 | true | 404 [en, ru], edit [en, ru], keeping upd, fixes | 1,842 | LAION-AI/Open-Assistant | 29,137 | Python | echo0x22 | Review this pull request change.
Language: TypeScript
File: website/src/pages/404.tsx
PR Title: 404 [en, ru], edit [en, ru], keeping upd, fixes
BEFORE CODE: import { Box, Button, Center, Link, Text, theme, useColorModeValue } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { AlertTria... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
const toast = useToast();
const handleCopyClick = async (event: MouseEvent) => {
event.stopPropagation();
toast.closeAll();
try {
await navigator.clipboard.writeText(String(children));
toast({
title: "Copied to clipboard",
status: "info",
duration: 2000,
is... | No issues found. | const toast = useToast();
const handleCopyClick = async (event: MouseEvent) => {
event.stopPropagation();
toast.closeAll();
try {
await navigator.clipboard.writeText(String(children));
toast({
title: "Copied to clipboard",
status: "info",
duration: 2000,
is... | @@ -0,0 +1,71 @@
+import { Button, Flex, useToast } from "@chakra-ui/react";
+import { Copy, Check } from "lucide-react";
+import { useState, MouseEvent } from "react";
+import { Prism as SyntaxHighlighter } from "react-syntax-highlighter";
+import { oneDark } from "react-syntax-highlighter/dist/cjs/styles/prism";
+
+e... | website/src/components/Messages/RenderedCodeblock.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Add button to copy code in markdown | 1,845 | LAION-AI/Open-Assistant | 29,137 | Python | Slowlife01 | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/RenderedCodeblock.tsx
PR Title: Add button to copy code in markdown
BEFORE CODE: const toast = useToast();
const handleCopyClick = async (event: MouseEvent) => {
event.stopPropagation();
toast.closeAll();
t... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { del } from "src/lib/api";
import { API_ROUTES } from "src/lib/routes";
import { mutate } from "swr";
import useSWRMutation from "swr/mutation";
export const useDeleteMessage = (messageId: string) => {
return useSWRMutation(API_ROUTES.ADMIN_DELETE_MESSAGE(messageId), del, {
onSuccess: async () => {
... | No issues found. | import { del } from "src/lib/api";
import { API_ROUTES } from "src/lib/routes";
import { mutate } from "swr";
import useSWRMutation from "swr/mutation";
export const useDeleteMessage = (messageId: string) => {
return useSWRMutation(API_ROUTES.ADMIN_DELETE_MESSAGE(messageId), del, {
onSuccess: async () => {
... | @@ -0,0 +1,17 @@
+import { del } from "src/lib/api";
+import { API_ROUTES } from "src/lib/routes";
+import { mutate } from "swr";
+import useSWRMutation from "swr/mutation";
+
+export const useDeleteMessage = (messageId: string) => {
+ return useSWRMutation(API_ROUTES.ADMIN_DELETE_MESSAGE(messageId), del, {
+ onSuc... | website/src/hooks/message/useDeleteMessage.ts | 0 | TypeScript | 1 | none | 16 | 18 | 18 | true | Add ability to Delete Messages From the List View in User Detail View… | 1,860 | LAION-AI/Open-Assistant | 29,137 | Python | johnflux | Review this pull request change.
Language: TypeScript
File: website/src/hooks/message/useDeleteMessage.ts
PR Title: Add ability to Delete Messages From the List View in User Detail View…
BEFORE CODE: import { del } from "src/lib/api";
import { API_ROUTES } from "src/lib/routes";
import { mutate } from "swr";
import... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { Button, CircularProgress, Flex, Spacer } from "@chakra-ui/react";
import { useTranslation } from "react-i18next";
import { get } from "src/lib/api";
import { FetchUserMessagesCursorResponse } from "src/types/Conversation";
import useSWRImmutable from "swr/immutable";
import { useCursorPagination } from "./Dat... | No issues found. | import { Button, CircularProgress, Flex, Spacer } from "@chakra-ui/react";
import { useTranslation } from "react-i18next";
import { get } from "src/lib/api";
import { FetchUserMessagesCursorResponse } from "src/types/Conversation";
import useSWRImmutable from "swr/immutable";
import { useCursorPagination } from "./Dat... | @@ -0,0 +1,47 @@
+import { Button, CircularProgress, Flex, Spacer } from "@chakra-ui/react";
+import { useTranslation } from "react-i18next";
+import { get } from "src/lib/api";
+import { FetchUserMessagesCursorResponse } from "src/types/Conversation";
+import useSWRImmutable from "swr/immutable";
+
+import { useCursor... | website/src/components/UserMessageConversation.tsx | 0 | TypeScript | 1 | none | 16 | 48 | 48 | true | Show all my messages | 1,864 | LAION-AI/Open-Assistant | 29,137 | Python | johnflux | Review this pull request change.
Language: TypeScript
File: website/src/components/UserMessageConversation.tsx
PR Title: Show all my messages
BEFORE CODE: import { Button, CircularProgress, Flex, Spacer } from "@chakra-ui/react";
import { useTranslation } from "react-i18next";
import { get } from "src/lib/api";
imp... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
@app.get("/auth/callback/discord", response_model=protocol.Token)
async def callback_discord(
code: str,
db: sqlmodel.Session = Depends(deps.create_session),
):
redirect_uri = f"{settings.api_root}/auth/callback/discord"
async with aiohttp.ClientSession(raise_for_status=True) as session:
# Ex... | No issues found. |
@app.get("/auth/callback/discord", response_model=protocol.Token)
async def callback_discord(
code: str,
db: sqlmodel.Session = Depends(deps.create_session),
):
redirect_uri = f"{settings.api_root}/auth/callback/discord"
async with aiohttp.ClientSession(raise_for_status=True) as session:
# Ex... | @@ -20,6 +20,16 @@
app = fastapi.FastAPI()
+@app.middleware("http")
+async def log_exceptions(request: fastapi.Request, call_next):
+ try:
+ response = await call_next(request)
+ except Exception:
+ logger.exception("Exception in request")
+ raise
+ return response
+
+
# add promethe... | inference/server/main.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Various fixes in inference | 1,865 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/server/main.py
PR Title: Various fixes in inference
BEFORE CODE:
@app.get("/auth/callback/discord", response_model=protocol.Token)
async def callback_discord(
code: str,
db: sqlmodel.Session = Depends(deps.create_session),
):
redirect_u... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
special_tokens: SpecialTokens = {}
TOKENIZER_CONFIGS = {
"galactica": TokenizerConfig(special_tokens=SpecialTokens("<pad>", "</s>")),
"GPT-JT": TokenizerConfig(special_tokens=SpecialTokens(sep_token="<|extratoken_100|>")),
"codegen": TokenizerConfig(special_tokens=SpecialTokens("<|endoftext|>", sep_to... | No issues found. | special_tokens: SpecialTokens = {}
TOKENIZER_CONFIGS = {
"galactica": TokenizerConfig(special_tokens=SpecialTokens("<pad>", "</s>")),
"GPT-JT": TokenizerConfig(special_tokens=SpecialTokens(sep_token="<|extratoken_100|>")),
"codegen": TokenizerConfig(special_tokens=SpecialTokens("<|endoftext|>", sep_to... | @@ -1,3 +1,4 @@
+import copy
import random
from distutils.util import strtobool
from pathlib import Path
@@ -12,14 +13,14 @@
from models import freeze_top_n_layers, get_specific_model
from sklearn.model_selection import train_test_split
from torch.utils.data import ConcatDataset, Subset
-from torch.utils.data.sam... | model/model_training/utils.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Distributed Sampler | 1,869 | LAION-AI/Open-Assistant | 29,137 | Python | sanagno | Review this pull request change.
Language: Python
File: model/model_training/utils.py
PR Title: Distributed Sampler
BEFORE CODE: special_tokens: SpecialTokens = {}
TOKENIZER_CONFIGS = {
"galactica": TokenizerConfig(special_tokens=SpecialTokens("<pad>", "</s>")),
"GPT-JT": TokenizerConfig(special_token... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { RefObject, useEffect, useState } from "react";
/**
* This is a custom hook that returns a ref that can be used to focus an element.
*
* Use like:
*
* ```
* const ref = useRef();
* useScrollToElementOnMount(ref);
* return <div ref={ref} />
* ```
*/
export const useScrollToElementOnMount ... | No issues found. | import { RefObject, useEffect, useState } from "react";
/**
* This is a custom hook that returns a ref that can be used to focus an element.
*
* Use like:
*
* ```
* const ref = useRef();
* useScrollToElementOnMount(ref);
* return <div ref={ref} />
* ```
*/
export const useScrollToElementOnMount ... | @@ -0,0 +1,24 @@
+import { RefObject, useEffect, useState } from "react";
+
+/**
+ * This is a custom hook that returns a ref that can be used to focus an element.
+ *
+ * Use like:
+ *
+ * ```
+ * const ref = useRef();
+ * useScrollToElementOnMount(ref);
+ * return <div ref={ref} />
+ * ```
+ */
+export c... | website/src/hooks/ui/useScrollToElementOnMount.ts | 0 | TypeScript | 1 | none | 16 | 25 | 25 | true | Scroll to message when loaded | 1,886 | LAION-AI/Open-Assistant | 29,137 | Python | johnflux | Review this pull request change.
Language: TypeScript
File: website/src/hooks/ui/useScrollToElementOnMount.ts
PR Title: Scroll to message when loaded
BEFORE CODE: import { RefObject, useEffect, useState } from "react";
/**
* This is a custom hook that returns a ref that can be used to focus an element.
*
* U... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { Button, Progress } from "@chakra-ui/react";
import Head from "next/head";
import { useTranslation } from "next-i18next";
import { useCallback, useMemo } from "react";
import { getDashboardLayout } from "src/components/Layout";
import { post } from "src/lib/api";
export { getDefaultStaticProps as getStaticProps... | No issues found. | import { Button, Progress } from "@chakra-ui/react";
import Head from "next/head";
import { useTranslation } from "next-i18next";
import { useCallback, useMemo } from "react";
import { getDashboardLayout } from "src/components/Layout";
import { post } from "src/lib/api";
export { getDefaultStaticProps as getStaticProps... | @@ -1,134 +1,46 @@
-import { Box, Button, Flex, Textarea, useColorModeValue } from "@chakra-ui/react";
+import { Button, Progress } from "@chakra-ui/react";
import Head from "next/head";
import { useTranslation } from "next-i18next";
-import { useCallback, useRef, useState } from "react";
+import { useCallback, useMe... | website/src/pages/chat.tsx | 0 | TypeScript | 1 | none | 16 | 47 | 47 | true | Add debug auth between Website & Inference | 1,893 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/pages/chat.tsx
PR Title: Add debug auth between Website & Inference
BEFORE CODE: import { Button, Progress } from "@chakra-ui/react";
import Head from "next/head";
import { useTranslation } from "next-i18next";
import { useCallback, useMemo } f... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import enum
from uuid import uuid4
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from oasst_shared.schemas import inference
from sqlmodel import Field, Relationship, SQLModel
class WorkerEventType(str, enum.Enum):
connect = "connect"
class DbWorkerComplianceCheck(SQLModel, table=True):
... | No issues found. | import enum
from uuid import uuid4
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from oasst_shared.schemas import inference
from sqlmodel import Field, Relationship, SQLModel
class WorkerEventType(str, enum.Enum):
connect = "connect"
class DbWorkerComplianceCheck(SQLModel, table=True):
... | @@ -0,0 +1,51 @@
+import datetime
+import enum
+from uuid import uuid4
+
+import sqlalchemy as sa
+import sqlalchemy.dialects.postgresql as pg
+from oasst_shared.schemas import inference
+from sqlmodel import Field, Relationship, SQLModel
+
+
+class WorkerEventType(str, enum.Enum):
+ connect = "connect"
+
+
+class D... | inference/server/oasst_inference_server/models/worker.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Store worker compliance check results, add initial scoring algorithm | 1,894 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/models/worker.py
PR Title: Store worker compliance check results, add initial scoring algorithm
BEFORE CODE: import enum
from uuid import uuid4
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from oa... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
"""Add score to message
Revision ID: 0f89c99ffdc8
Revises: a2865908a537
Create Date: 2023-03-01 14:42:18.577510
"""
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision = "0f89c99ffdc8"
down_revision = "a2865908a537"
branch_labels = None
depends_on = None
def upgrade() -... | No issues found. | """Add score to message
Revision ID: 0f89c99ffdc8
Revises: a2865908a537
Create Date: 2023-03-01 14:42:18.577510
"""
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision = "0f89c99ffdc8"
down_revision = "a2865908a537"
branch_labels = None
depends_on = None
def upgrade() -... | @@ -0,0 +1,38 @@
+"""Add score to message
+
+Revision ID: 0f89c99ffdc8
+Revises: a2865908a537
+Create Date: 2023-03-01 14:42:18.577510
+
+"""
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "0f89c99ffdc8"
+down_revision = "a2865908a537"
+branch_labels = None
+dep... | inference/server/alembic/versions/2023_03_01_1442-0f89c99ffdc8_add_score_to_message.py | 0 | Python | 1 | none | 16 | 39 | 39 | true | refactor inference message score | 1,907 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: Python
File: inference/server/alembic/versions/2023_03_01_1442-0f89c99ffdc8_add_score_to_message.py
PR Title: refactor inference message score
BEFORE CODE: """Add score to message
Revision ID: 0f89c99ffdc8
Revises: a2865908a537
Create Date: 2023-03-01 14:42:18.577510
"""... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
userEmoji={emojiState.user_emojis}
onLabel={showLabelPopup}
onReport={showReportPopup}
message={message}
/>
<LabelMessagePopup message={message} show={labelPopupOpen} onClose={closeLabelPopup} />
<ReportPopup messageId={message.id} show={repo... | No issues found. | userEmoji={emojiState.user_emojis}
onLabel={showLabelPopup}
onReport={showReportPopup}
message={message}
/>
<LabelMessagePopup message={message} show={labelPopupOpen} onClose={closeLabelPopup} />
<ReportPopup messageId={message.id} show={repo... | @@ -1,9 +1,6 @@
import {
- Avatar,
Badge,
- Box,
Flex,
- HStack,
Menu,
MenuButton,
MenuDivider,
@@ -13,7 +10,6 @@ import {
Portal,
SimpleGrid,
Tooltip,
- useBreakpointValue,
useColorModeValue,
useDisclosure,
useToast,
@@ -34,119 +30,84 @@ import {
import NextLink from "next/link";... | website/src/components/Messages/MessageTableEntry.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Chat vote UI | 1,913 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/MessageTableEntry.tsx
PR Title: Chat vote UI
BEFORE CODE: userEmoji={emojiState.user_emojis}
onLabel={showLabelPopup}
onReport={showReportPopup}
message={message}
/>
... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
if event.user_id is not None:
event.user_id = anonymizer.anonymize("user", event.user_id)
case "ranking":
event: ExportMessageEventRanking = event
if event.user_id is not None:
... | No issues found. | if event.user_id is not None:
event.user_id = anonymizer.anonymize("user", event.user_id)
case "ranking":
event: ExportMessageEventRanking = event
if event.user_id is not None:
... | @@ -2,22 +2,96 @@
import contextlib
import gzip
+import hashlib
import json
import sys
+import uuid
from collections import defaultdict
from typing import Iterable, Optional, TextIO
-from uuid import UUID
from fastapi.encoders import jsonable_encoder
from oasst_backend.models import Message
from oasst_back... | backend/oasst_backend/utils/tree_export.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Extended export script | 1,954 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: backend/oasst_backend/utils/tree_export.py
PR Title: Extended export script
BEFORE CODE: if event.user_id is not None:
event.user_id = anonymizer.anonymize("user", event.user_id)
... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { Box, Divider, Text, useColorMode } from "@chakra-ui/react";
import { useSession } from "next-auth/react";
import { useTranslation } from "next-i18next";
export function WelcomeCard() {
const { colorMode } = useColorMode();
const backgroundColor = colorMode === "light" ? "white" : "gray.700";
const title... | No issues found. | import { Box, Divider, Text, useColorMode } from "@chakra-ui/react";
import { useSession } from "next-auth/react";
import { useTranslation } from "next-i18next";
export function WelcomeCard() {
const { colorMode } = useColorMode();
const backgroundColor = colorMode === "light" ? "white" : "gray.700";
const title... | @@ -10,32 +10,25 @@ export function WelcomeCard() {
const { data: session } = useSession();
const { t } = useTranslation("dashboard");
- if (!session) {
- return <></>;
+ const isNew = session?.user?.isNew;
+ if (!isNew) {
+ return null;
}
- if (session && session.user && session.user.isNew)
- r... | website/src/components/Dashboard/WelcomeCard.tsx | 0 | TypeScript | 1 | none | 16 | 35 | 35 | true | Fix some null checks | 1,963 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Dashboard/WelcomeCard.tsx
PR Title: Fix some null checks
BEFORE CODE: import { Box, Divider, Text, useColorMode } from "@chakra-ui/react";
import { useSession } from "next-auth/react";
import { useTranslation } from "next-i18next";
... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { HStack } from "@chakra-ui/react";
import { Icon } from "@chakra-ui/react";
import { Star } from "lucide-react";
import { useTranslation } from "react-i18next";
import { useUserScore } from "src/hooks/ui/useUserScore";
export const UserScore = () => {
const score = useUserScore();
const { t } = useTranslat... | No issues found. | import { HStack } from "@chakra-ui/react";
import { Icon } from "@chakra-ui/react";
import { Star } from "lucide-react";
import { useTranslation } from "react-i18next";
import { useUserScore } from "src/hooks/ui/useUserScore";
export const UserScore = () => {
const score = useUserScore();
const { t } = useTranslat... | @@ -0,0 +1,21 @@
+import { HStack } from "@chakra-ui/react";
+import { Icon } from "@chakra-ui/react";
+import { Star } from "lucide-react";
+import { useTranslation } from "react-i18next";
+import { useUserScore } from "src/hooks/ui/useUserScore";
+
+export const UserScore = () => {
+ const score = useUserScore();
+ ... | website/src/components/Header/UserScore.tsx | 0 | TypeScript | 1 | none | 16 | 22 | 22 | true | Show user score in header | 1,969 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Header/UserScore.tsx
PR Title: Show user score in header
BEFORE CODE: import { HStack } from "@chakra-ui/react";
import { Icon } from "@chakra-ui/react";
import { Star } from "lucide-react";
import { useTranslation } from "react-i18n... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import { Stack, Text } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { MessageTableEntry } from "src/components/Messages/MessageTableEntry";
import { Message } from "src/types/Conversation";
export interface MessageConversationProps {
messages: Message[];
enableLink?: boolean;
hi... | No issues found. | import { Stack, Text } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { MessageTableEntry } from "src/components/Messages/MessageTableEntry";
import { Message } from "src/types/Conversation";
export interface MessageConversationProps {
messages: Message[];
enableLink?: boolean;
hi... | @@ -7,9 +7,15 @@ export interface MessageConversationProps {
messages: Message[];
enableLink?: boolean;
highlightLastMessage?: boolean;
+ showCreatedDate?: boolean;
}
-export function MessageConversation({ messages, enableLink, highlightLastMessage }: MessageConversationProps) {
+export function MessageCon... | website/src/components/Messages/MessageConversation.tsx | 0 | TypeScript | 1 | none | 16 | 38 | 38 | true | show message create date | 1,978 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/MessageConversation.tsx
PR Title: show message create date
BEFORE CODE: import { Stack, Text } from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
import { MessageTableEntry } from "src/components/Message... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
num_processes=self.args.world_size,
process_index=self.args.process_index,
)
return DataLoader(
train_dataset,
batch_size=self.args.per_device_train_batch_size,
collate_fn=data_collator,
n... | No issues found. | num_processes=self.args.world_size,
process_index=self.args.process_index,
)
return DataLoader(
train_dataset,
batch_size=self.args.per_device_train_batch_size,
collate_fn=data_collator,
n... | @@ -214,10 +214,18 @@ def argument_parsing(notebook=False, notebook_args=None):
train_collate_fn = DialogueDataCollator(
tokenizer,
max_length=training_conf.max_length,
+ random_offset_probability=training_conf.random_offset_probability,
+ label_masking=training_conf.label_masking,
... | model/model_training/trainer_sft.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Add SFT params random_offset_probability, label_masking | 2,005 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/trainer_sft.py
PR Title: Add SFT params random_offset_probability, label_masking
BEFORE CODE: num_processes=self.args.world_size,
process_index=self.args.process_index,
)
... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
def get_message(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: ApiClient = Depends(deps.get_api_client),
db: Session = Depends(deps.get_db),
):
"""
Get a message by its internal ID.
"""
pr = PromptRepository(db, api_client, ... | No issues found. | def get_message(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: ApiClient = Depends(deps.get_api_client),
db: Session = Depends(deps.get_db),
):
"""
Get a message by its internal ID.
"""
pr = PromptRepository(db, api_client, ... | @@ -70,6 +70,7 @@ def get_messages_cursor(
max_count: Optional[int] = Query(10, gt=0, le=1000),
desc: Optional[bool] = False,
lang: Optional[str] = None,
+ include_user: Optional[bool] = None,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: ApiClient = De... | backend/oasst_backend/api/v1/messages.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | add `include_user` params in /messages/cursor` endpoint | 2,021 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: Python
File: backend/oasst_backend/api/v1/messages.py
PR Title: add `include_user` params in /messages/cursor` endpoint
BEFORE CODE: def get_message(
*,
message_id: UUID,
frontend_user: deps.FrontendUserId = Depends(deps.get_frontend_user_id),
api_client: A... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
"""
These are in the form of 'INSTRUCTION', 'RESPONSE'
"""
from custom_datasets.formatting import format_pair
from datasets import load_dataset
from torch.utils.data import Dataset
INSTRUCTION_DATASETS = {
# Note humaneval_mbpp_codegen_qa returns a code string that we would want to at least wrap in ``` marks`
... | No issues found. | """
These are in the form of 'INSTRUCTION', 'RESPONSE'
"""
from custom_datasets.formatting import format_pair
from datasets import load_dataset
from torch.utils.data import Dataset
INSTRUCTION_DATASETS = {
# Note humaneval_mbpp_codegen_qa returns a code string that we would want to at least wrap in ``` marks`
... | @@ -0,0 +1,36 @@
+"""
+ These are in the form of 'INSTRUCTION', 'RESPONSE'
+"""
+from custom_datasets.formatting import format_pair
+from datasets import load_dataset
+from torch.utils.data import Dataset
+
+INSTRUCTION_DATASETS = {
+ # Note humaneval_mbpp_codegen_qa returns a code string that we would want to at... | model/model_training/custom_datasets/instruction.py | 0 | Python | 1 | none | 16 | 37 | 37 | true | Add option to tokenize numbers per digit | 2,024 | LAION-AI/Open-Assistant | 29,137 | Python | johnflux | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/instruction.py
PR Title: Add option to tokenize numbers per digit
BEFORE CODE: """
These are in the form of 'INSTRUCTION', 'RESPONSE'
"""
from custom_datasets.formatting import format_pair
from datasets import load_data... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
import Image from "next/image";
import { useEffect, useState } from "react";
export function UserAvatar({ displayName, avatarUrl }: { displayName: string; avatarUrl?: string }) {
const diceBearURL = `https://api.dicebear.com/5.x/initials/png?seed=${displayName}&radius=50&backgroundType=gradientLinear`;
const [src,... | No issues found. | import Image from "next/image";
import { useEffect, useState } from "react";
export function UserAvatar({ displayName, avatarUrl }: { displayName: string; avatarUrl?: string }) {
const diceBearURL = `https://api.dicebear.com/5.x/initials/png?seed=${displayName}&radius=50&backgroundType=gradientLinear`;
const [src,... | @@ -1,26 +1,21 @@
import Image from "next/image";
-import { useState } from "react";
-export function UserAvatar(options: { displayName: string; avatarUrl?: string }) {
- const { displayName, avatarUrl } = options;
- const [src, setSrc] = useState(
- avatarUrl
- ? avatarUrl
- : `https://api.dicebear.com... | website/src/components/UserAvatar.tsx | 0 | TypeScript | 1 | none | 16 | 22 | 22 | true | Show correct user avatars on leaderboard | 2,039 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/UserAvatar.tsx
PR Title: Show correct user avatars on leaderboard
BEFORE CODE: import Image from "next/image";
import { useEffect, useState } from "react";
export function UserAvatar({ displayName, avatarUrl }: { displayName: string;... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
self.data = [json.loads(lines[index].strip()) for index in indices[start_index:end_index]]
def __len__(self):
return len(self.data)
def __getitem__(self, index):
# Sample randomly from replies
prompt = self.data[index]["prompt"]
pairs = []
while True:
... | No issues found. |
self.data = [json.loads(lines[index].strip()) for index in indices[start_index:end_index]]
def __len__(self):
return len(self.data)
def __getitem__(self, index):
# Sample randomly from replies
prompt = self.data[index]["prompt"]
pairs = []
while True:
... | @@ -6,7 +6,6 @@
from functools import reduce
import numpy as np
-from custom_datasets.formatting import QA_SPECIAL_TOKENS, format_pair
from torch.utils.data import Dataset
@@ -68,56 +67,56 @@ def __getitem__(self, index):
if self.mode == "sft":
# return all dialogue history
- ... | model/model_training/custom_datasets/prompt_dialogue.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Choice to add global system-prefix to the assistant during changes | 2,053 | LAION-AI/Open-Assistant | 29,137 | Python | sanagno | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/prompt_dialogue.py
PR Title: Choice to add global system-prefix to the assistant during changes
BEFORE CODE:
self.data = [json.loads(lines[index].strip()) for index in indices[start_index:end_index]]
def __len... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
const { register, handleSubmit } = useForm<UserForm>({
defaultValues: user,
});
return (
<>
<Head>
<title>Manage Users - Open Assistant</title>
<meta
name="description"
content="Conversational AI for everyone. An open source project to create a chat enabled GPT ... | No issues found. |
const { register, handleSubmit } = useForm<UserForm>({
defaultValues: user,
});
return (
<>
<Head>
<title>Manage Users - Open Assistant</title>
<meta
name="description"
content="Conversational AI for everyone. An open source project to create a chat enabled GPT ... | @@ -10,7 +10,6 @@ import {
CardBody,
CardHeader,
Checkbox,
- CircularProgress,
FormControl,
FormLabel,
Input,
@@ -22,19 +21,16 @@ import Head from "next/head";
import { serverSideTranslations } from "next-i18next/serverSideTranslations";
import { useForm } from "react-hook-form";
import { AdminAre... | website/src/pages/admin/manage_user/[id].tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Admin message list | 2,057 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/pages/admin/manage_user/[id].tsx
PR Title: Admin message list
BEFORE CODE:
const { register, handleSubmit } = useForm<UserForm>({
defaultValues: user,
});
return (
<>
<Head>
<title>Manage Users - Open Assistant</ti... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
return (
<Box display="-webkit-box" wordBreak="break-all" whiteSpace="pre-wrap" w="md">
<Avatar
size="xs"
mr="2"
src={`${row.original.is_assistant ? "/images/logos/logo.png" : "/images/temp-avatars/av1.jpg"}`}
></Avatar>
... | No issues found. | return (
<Box display="-webkit-box" wordBreak="break-all" whiteSpace="pre-wrap" w="md">
<Avatar
size="xs"
mr="2"
src={`${row.original.is_assistant ? "/images/logos/logo.png" : "/images/temp-avatars/av1.jpg"}`}
></Avatar>
... | @@ -14,7 +14,7 @@ import {
} from "@chakra-ui/react";
import { createColumnHelper } from "@tanstack/table-core";
import { formatDistanceToNow, formatISO9075 } from "date-fns";
-import { Eye, Trash } from "lucide-react";
+import { Eye, RotateCw, Trash } from "lucide-react";
import NextLink from "next/link";
import ... | website/src/components/Messages/AdminMessageTable.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | [#2077] Possbility to revert message deletion | 2,079 | LAION-AI/Open-Assistant | 29,137 | Python | Shadowner | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/AdminMessageTable.tsx
PR Title: [#2077] Possbility to revert message deletion
BEFORE CODE: return (
<Box display="-webkit-box" wordBreak="break-all" whiteSpace="pre-wrap" w="md">
<Avatar
... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
collate_fn=data_collator,
drop_last=self.args.dataloader_drop_last,
num_workers=self.args.dataloader_num_workers,
pin_memory=self.args.dataloader_pin_memory,
worker_init_fn=seed_worker,
)
return dataloader
def argument_parsing(notebook=False,... | No issues found. | collate_fn=data_collator,
drop_last=self.args.dataloader_drop_last,
num_workers=self.args.dataloader_num_workers,
pin_memory=self.args.dataloader_pin_memory,
worker_init_fn=seed_worker,
)
return dataloader
def argument_parsing(notebook=False,... | @@ -187,8 +187,6 @@ def argument_parsing(notebook=False, notebook_args=None):
else:
args, remaining = parser.parse_known_args()
- print(args)
-
# Config from YAML
conf = {}
configs = read_yamls("./configs")
@@ -222,8 +220,12 @@ def argument_parsing(notebook=False, notebook_args=None):
... | model/model_training/trainer_sft.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Add minimath support | 2,094 | LAION-AI/Open-Assistant | 29,137 | Python | johnflux | Review this pull request change.
Language: Python
File: model/model_training/trainer_sft.py
PR Title: Add minimath support
BEFORE CODE: collate_fn=data_collator,
drop_last=self.args.dataloader_drop_last,
num_workers=self.args.dataloader_num_workers,
pin_memory=self.ar... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
justifyContent="space-between"
alignItems="center"
gap="6"
p="8"
pb={["14", "8"]}
w="full"
mx="auto"
maxWidth="7xl"
>
<Flex alignItems="center">
<Box pr="2">
<Link href="/" aria-label="Home">
... | No issues found. | justifyContent="space-between"
alignItems="center"
gap="6"
p="8"
pb={["14", "8"]}
w="full"
mx="auto"
maxWidth="7xl"
>
<Flex alignItems="center">
<Box pr="2">
<Link href="/" aria-label="Home">
... | @@ -63,6 +63,7 @@ export function Footer() {
<Text fontWeight="bold" color={textColor}>
{t("about")}
</Text>
+ <FooterLink href="/team" label={t("who_are_we")} />
<FooterLink href="https://projects.laion.ai/Open-Assistant" label={t("do... | website/src/components/Footer.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Add Team page | 2,103 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Footer.tsx
PR Title: Add Team page
BEFORE CODE: justifyContent="space-between"
alignItems="center"
gap="6"
p="8"
pb={["14", "8"]}
w="full"
mx="auto"
maxW... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
temperature: 1,
top_k: 50 / 32000,
top_p: 0.95,
typical_p: 0.5,
};
const parameterLabel: Record<keyof typeof customPresetDefaultValue, string> = {
max_new_tokens: "Max new tokens",
top_k: "Top K",
top_p: "Top P",
temperature: "Temperature",
repetition_penalty: "Repetition Penalty",
typical_p: "Typi... | No issues found. | temperature: 1,
top_k: 50 / 32000,
top_p: 0.95,
typical_p: 0.5,
};
const parameterLabel: Record<keyof typeof customPresetDefaultValue, string> = {
max_new_tokens: "Max new tokens",
top_k: "Top K",
top_p: "Top P",
temperature: "Temperature",
repetition_penalty: "Repetition Penalty",
typical_p: "Typi... | @@ -0,0 +1,269 @@
+import {
+ Drawer,
+ DrawerBody,
+ DrawerCloseButton,
+ DrawerContent,
+ DrawerHeader,
+ DrawerOverlay,
+ Flex,
+ FormControl,
+ FormLabel,
+ IconButton,
+ NumberDecrementStepper,
+ NumberIncrementStepper,
+ NumberInput,
+ NumberInputField,
+ NumberInputStepper,
+ Portal,
+ Select,
+... | website/src/components/Chat/ChatConfigDrawer.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Chat config UI | 2,115 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConfigDrawer.tsx
PR Title: Chat config UI
BEFORE CODE: temperature: 1,
top_k: 50 / 32000,
top_p: 0.95,
typical_p: 0.5,
};
const parameterLabel: Record<keyof typeof customPresetDefaultValue, string> = {
max_new_tok... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
<stop offset="1" stopColor={baseRingColor} stopOpacity="0" />
</linearGradient>
</defs>
<path
opacity=".2"
d="M1 279C1 125.465 125.465 1 279 1s278 124.465 278 278-124.465 278-278 278S1 432.535 1 279Z"
stroke={baseRingColor}
/>
<path d="M1 279C1 125.465 1... | No issues found. | <stop offset="1" stopColor={baseRingColor} stopOpacity="0" />
</linearGradient>
</defs>
<path
opacity=".2"
d="M1 279C1 125.465 125.465 1 279 1s278 124.465 278 278-124.465 278-278 278S1 432.535 1 279Z"
stroke={baseRingColor}
/>
<path d="M1 279C1 125.465 1... | @@ -1,5 +1,5 @@
import { Box, Link, Text, useColorMode } from "@chakra-ui/react";
-import { Github } from "lucide-react";
+import { Github, Users } from "lucide-react";
import { useTranslation } from "next-i18next";
import { useId } from "react";
@@ -62,14 +62,16 @@ export function CallToAction() {
<Tex... | website/src/components/CallToAction.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Updated team page to fit visually more the the rest of the pages, added a button to the call-to-action section which links to the team page. | 2,126 | LAION-AI/Open-Assistant | 29,137 | Python | theopfr | Review this pull request change.
Language: TypeScript
File: website/src/components/CallToAction.tsx
PR Title: Updated team page to fit visually more the the rest of the pages, added a button to the call-to-action section which links to the team page.
BEFORE CODE: <stop offset="1" stopColor={baseRingColor}... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
origin: string;
};
}>(`/api/admin/messages/${messageId}/tree`, get);
return (
<>
<Head>
<title>Open Assistant</title>
</Head>
<AdminArea>
{isLoading && <CircularProgress isIndeterminate></CircularProgress>}
{error && "Unable to load message tree"}
{da... | No issues found. | origin: string;
};
}>(`/api/admin/messages/${messageId}/tree`, get);
return (
<>
<Head>
<title>Open Assistant</title>
</Head>
<AdminArea>
{isLoading && <CircularProgress isIndeterminate></CircularProgress>}
{error && "Unable to load message tree"}
{da... | @@ -1,4 +1,19 @@
-import { Card, CardBody, CardHeader, CircularProgress, Grid } from "@chakra-ui/react";
+import {
+ Card,
+ CardBody,
+ CardHeader,
+ CircularProgress,
+ Grid,
+ Text,
+ TableContainer,
+ Table,
+ TableCaption,
+ Thead,
+ Tr,
+ Th,
+ Tbody,
+ Td,
+} from "@chakra-ui/react";
import { GetS... | website/src/pages/admin/messages/[id].tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Display message tree state admin | 2,133 | LAION-AI/Open-Assistant | 29,137 | Python | Rulios | Review this pull request change.
Language: TypeScript
File: website/src/pages/admin/messages/[id].tsx
PR Title: Display message tree state admin
BEFORE CODE: origin: string;
};
}>(`/api/admin/messages/${messageId}/tree`, get);
return (
<>
<Head>
<title>Open Assistant</title>
<... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
};
}
return {
props: {
...(await serverSideTranslations(locale)),
},
};
};
const Chat = () => {
const { t } = useTranslation(["common", "chat"]);
const router = useRouter();
const { data } = useSWR<GetChatsResponse>("/api/chat", get, { revalidateOnFocus: true });
const { trigger: newC... | No issues found. | };
}
return {
props: {
...(await serverSideTranslations(locale)),
},
};
};
const Chat = () => {
const { t } = useTranslation(["common", "chat"]);
const router = useRouter();
const { data } = useSWR<GetChatsResponse>("/api/chat", get, { revalidateOnFocus: true });
const { trigger: newC... | @@ -9,24 +9,25 @@ import { useCallback, useMemo } from "react";
import { getDashboardLayout } from "src/components/Layout";
import { SurveyCard } from "src/components/Survey/SurveyCard";
import { get, post } from "src/lib/api";
+import { isChatEnabled } from "src/lib/chat_enabled";
import { GetChatsResponse } from ... | website/src/pages/chat/index.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Added ENABLE_CHAT to chat id page and docker-compose | 2,137 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: TypeScript
File: website/src/pages/chat/index.tsx
PR Title: Added ENABLE_CHAT to chat id page and docker-compose
BEFORE CODE: };
}
return {
props: {
...(await serverSideTranslations(locale)),
},
};
};
const Chat = () => {
const { t } = useTransla... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
return inference.compat_hash(model_name=self.work_parameters.model_name)
class CreateMessageResponse(pydantic.BaseModel):
prompter_message: inference.MessageRead
assistant_message: inference.MessageRead
class PendingResponseEvent(pydantic.BaseModel):
event_type: Literal["pending"] = "pending"
... | No issues found. | return inference.compat_hash(model_name=self.work_parameters.model_name)
class CreateMessageResponse(pydantic.BaseModel):
prompter_message: inference.MessageRead
assistant_message: inference.MessageRead
class PendingResponseEvent(pydantic.BaseModel):
event_type: Literal["pending"] = "pending"
... | @@ -70,3 +70,9 @@ class ChatRead(ChatListRead):
class ListChatsResponse(pydantic.BaseModel):
chats: list[ChatListRead]
+
+
+class MessageCancelledException(Exception):
+ def __init__(self, message_id: str):
+ super().__init__(f"Message {message_id} was cancelled")
+ self.message_id = message_id | inference/server/oasst_inference_server/schemas/chat.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Cancelling pending chat messages of user on submission of a new message | 2,141 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/schemas/chat.py
PR Title: Cancelling pending chat messages of user on submission of a new message
BEFORE CODE: return inference.compat_hash(model_name=self.work_parameters.model_name)
class CreateMessageRespons... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
if eos_token:
parameters.stop.append(eos_token)
return prompt, parameters
def handle_work_request(
ws: websocket.WebSocket,
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
):
prompt, parameters = make_prompt_and_parameters(tokenizer, work_request)
... | No issues found. | if eos_token:
parameters.stop.append(eos_token)
return prompt, parameters
def handle_work_request(
ws: websocket.WebSocket,
tokenizer: transformers.PreTrainedTokenizer,
work_request: inference.WorkRequest,
):
prompt, parameters = make_prompt_and_parameters(tokenizer, work_request)
... | @@ -85,12 +85,15 @@ def handle_work_request(
)
if settings.model_id == "_lorem":
stream_events = utils.lorem_events(parameters.seed)
- elif "llama" in settings.model_id:
- prompt = truncate_prompt(tokenizer, parameters, prompt)
- stream_events = get_hf_stream_events(stream_request)
+... | inference/worker/work.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Llama inference | 2,144 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/worker/work.py
PR Title: Llama inference
BEFORE CODE: if eos_token:
parameters.stop.append(eos_token)
return prompt, parameters
def handle_work_request(
ws: websocket.WebSocket,
tokenizer: transformers.PreTrainedTokenizer,
... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
class ModelInfo(pydantic.BaseModel):
name: str
description: str = ""
parameter_configs: list[ParameterConfig] = []
DEFAULT_PARAMETER_CONFIGS = [
ParameterConfig(
name="k50",
description="Top-k sampling with k=50",
work_parameters=inference.WorkParametersInput(
mode... | No issues found. |
class ModelInfo(pydantic.BaseModel):
name: str
description: str = ""
parameter_configs: list[ParameterConfig] = []
DEFAULT_PARAMETER_CONFIGS = [
ParameterConfig(
name="k50",
description="Top-k sampling with k=50",
work_parameters=inference.WorkParametersInput(
mode... | @@ -20,29 +20,60 @@ class ModelInfo(pydantic.BaseModel):
parameter_configs: list[ParameterConfig] = []
+DEFAULT_PARAMETER_CONFIGS = [
+ ParameterConfig(
+ name="k50",
+ description="Top-k sampling with k=50",
+ work_parameters=inference.WorkParametersInput(
+ model_name="_mod... | inference/server/oasst_inference_server/routes/configs.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Added param configs from eval to pass to the frontend | 2,163 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/configs.py
PR Title: Added param configs from eval to pass to the frontend
BEFORE CODE:
class ModelInfo(pydantic.BaseModel):
name: str
description: str = ""
parameter_configs: list[ParameterConfig] = ... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
items: SideMenuItem[];
}
export const SIDE_MENU_WIDTH = {
MD: `100px`,
LG: `280px`,
};
export function SideMenu({ items }: SideMenuProps) {
const router = useRouter();
const { t } = useTranslation();
return (
<Card
position={{ base: "relative", md: "fixed" }}
p="4"
width={{ base: "1... | No issues found. | items: SideMenuItem[];
}
export const SIDE_MENU_WIDTH = {
MD: `100px`,
LG: `280px`,
};
export function SideMenu({ items }: SideMenuProps) {
const router = useRouter();
const { t } = useTranslation();
return (
<Card
position={{ base: "relative", md: "fixed" }}
p="4"
width={{ base: "1... | @@ -1,75 +1,84 @@
-import { Button, Card, Text, Tooltip } from "@chakra-ui/react";
+import { Box, Button, Card, Text, Tooltip } from "@chakra-ui/react";
import { LucideIcon } from "lucide-react";
import Link from "next/link";
import { useRouter } from "next/router";
import { useTranslation } from "next-i18next";
i... | website/src/components/SideMenu.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | rework layout | 2,172 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/SideMenu.tsx
PR Title: rework layout
BEFORE CODE: items: SideMenuItem[];
}
export const SIDE_MENU_WIDTH = {
MD: `100px`,
LG: `280px`,
};
export function SideMenu({ items }: SideMenuProps) {
const router = useRouter();
con... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. | |
// error: "/auth/error", -Will be used later
},
callbacks: {
/**
* Ensure we propagate the user's role when creating the session from the
* token.
*/
async session({ session, token }) {
session.user.role = token.role;
session.user.isNew = token.isNew;
session.user.name =... | No issues found. | // error: "/auth/error", -Will be used later
},
callbacks: {
/**
* Ensure we propagate the user's role when creating the session from the
* token.
*/
async session({ session, token }) {
session.user.role = token.role;
session.user.isNew = token.isNew;
session.user.name =... | @@ -1,5 +1,6 @@
import { PrismaAdapter } from "@next-auth/prisma-adapter";
import { boolean } from "boolean";
+import Cookies from "cookies";
import { generateUsername } from "friendly-username-generator";
import { NextApiRequest, NextApiResponse } from "next";
import type { AuthOptions } from "next-auth";
@@ -11,... | website/src/pages/api/auth/[...nextauth].ts | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Login to inference through website | 2,176 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/pages/api/auth/[...nextauth].ts
PR Title: Login to inference through website
BEFORE CODE: // error: "/auth/error", -Will be used later
},
callbacks: {
/**
* Ensure we propagate the user's role when creating the session from the... | No issues found. | You are an expert software engineer and code reviewer. Review pull requests carefully for correctness, bugs, security issues, performance problems, maintainability, and unnecessary complexity. Only raise actionable issues supported by the code. Explain why an issue matters and suggest a concrete fix when appropriate. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.