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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
from argparse import Namespace
import model_training.models.reward_model # noqa: F401
from model_training.models.reward_model import GPTNeoXRewardModel
from model_training.utils import get_tokenizer
from transformers import AutoModelForSequenceClassification, AutoTokenizer
def test_convert_model(
model_name: st... | No issues found. | from argparse import Namespace
import model_training.models.reward_model # noqa: F401
from model_training.models.reward_model import GPTNeoXRewardModel
from model_training.utils import get_tokenizer
from transformers import AutoModelForSequenceClassification, AutoTokenizer
def test_convert_model(
model_name: st... | @@ -0,0 +1,40 @@
+from argparse import Namespace
+
+import model_training.models.reward_model # noqa: F401
+from model_training.models.reward_model import GPTNeoXRewardModel
+from model_training.utils import get_tokenizer
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+
+def test_convert... | model/model_training/tests/test_rm_loading.py | 0 | Python | 1 | none | 16 | 41 | 41 | true | Add GPTNeoXRewardModel | 2,182 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/tests/test_rm_loading.py
PR Title: Add GPTNeoXRewardModel
BEFORE CODE: from argparse import Namespace
import model_training.models.reward_model # noqa: F401
from model_training.models.reward_model import GPTNeoXRewardModel
from model_tra... | 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. | |
generated_ids = []
decoded_text = ""
for stream_response in stream_events:
if stream_response.is_error:
logger.error(f"Error from inference server: {stream_response.error}")
utils.send_response(
ws,
inference.ErrorResponse(
... | No issues found. | generated_ids = []
decoded_text = ""
for stream_response in stream_events:
if stream_response.is_error:
logger.error(f"Error from inference server: {stream_response.error}")
utils.send_response(
ws,
inference.ErrorResponse(
... | @@ -1,4 +1,5 @@
import threading
+from concurrent import futures
import interface
import requests
@@ -22,8 +23,8 @@ def truncate_prompt(
with tokenizer_lock:
ids = tokenizer.encode(prompt)
- max_input_length = worker_config.model_max_input_length
- max_total_tokens = worker_config.model_max_t... | inference/worker/work.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Introduce model configs to abstract pairings of models and hardware | 2,194 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/worker/work.py
PR Title: Introduce model configs to abstract pairings of models and hardware
BEFORE CODE: generated_ids = []
decoded_text = ""
for stream_response in stream_events:
if stream_response.is_error:
logger.e... | 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 customPresetDefaultValue: SamplingParameters = {
max_new_tokens: 256,
repetition_penalty: 1.2,
temperature: 1,
top_k: 50,
top_p: 0.95,
typical_p: 0.5,
};
const parameterLabel: Record<keyof SamplingParameters, string> = {
max_new_tokens: "Max new tokens",
top_k: "Top K",
top_p: "Top P",
temper... | No issues found. | const customPresetDefaultValue: SamplingParameters = {
max_new_tokens: 256,
repetition_penalty: 1.2,
temperature: 1,
top_k: 50,
top_p: 0.95,
typical_p: 0.5,
};
const parameterLabel: Record<keyof SamplingParameters, string> = {
max_new_tokens: "Max new tokens",
top_k: "Top K",
top_p: "Top P",
temper... | @@ -28,8 +28,7 @@ import { Settings } from "lucide-react";
import { useTranslation } from "next-i18next";
import { ChangeEvent, useCallback, useState } from "react";
import { Controller, useFormContext, useWatch } from "react-hook-form";
-import { WorkParametersInput } from "src/types/Chat";
-import { StrictOmit } f... | website/src/components/Chat/ChatConfigDrawer.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Use new inference model config / API | 2,208 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConfigDrawer.tsx
PR Title: Use new inference model config / API
BEFORE CODE: const customPresetDefaultValue: SamplingParameters = {
max_new_tokens: 256,
repetition_penalty: 1.2,
temperature: 1,
top_k: 50,
top_p: 0.... | 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. | |
tokenizer: transformers.PreTrainedTokenizer = transformers.AutoTokenizer.from_pretrained(model_config.model_id)
while True:
try:
if not model_config.is_lorem:
utils.wait_for_inference_server(settings.inference_server_url)
if settings.perform_oom_test:
... | No issues found. | tokenizer: transformers.PreTrainedTokenizer = transformers.AutoTokenizer.from_pretrained(model_config.model_id)
while True:
try:
if not model_config.is_lorem:
utils.wait_for_inference_server(settings.inference_server_url)
if settings.perform_oom_test:
... | @@ -85,7 +85,12 @@ def main():
)
ftrs.append(ftr)
case "ping":
- utils.send_response(ws, inference.PongResponse(request_id=worker_request.id))
+ utils.send_respons... | inference/worker/__main__.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Create worker metrics manually for more control | 2,229 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/worker/__main__.py
PR Title: Create worker metrics manually for more control
BEFORE CODE: tokenizer: transformers.PreTrainedTokenizer = transformers.AutoTokenizer.from_pretrained(model_config.model_id)
while True:
try:
... | 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. | |
ucr: UserChatRepository
async with deps.manual_user_chat_repository(user_id) as ucr:
work_parameters = inference.WorkParameters(
model_config=model_config,
sampling_parameters=request.sampling_parameters,
)
assistant_message = await ucr... | No issues found. | ucr: UserChatRepository
async with deps.manual_user_chat_repository(user_id) as ucr:
work_parameters = inference.WorkParameters(
model_config=model_config,
sampling_parameters=request.sampling_parameters,
)
assistant_message = await ucr... | @@ -2,11 +2,10 @@
import pydantic
from fastapi import Depends
from loguru import logger
-from oasst_inference_server import auth, deps, models, queueing
+from oasst_inference_server import auth, chat_utils, deps, models, queueing
from oasst_inference_server.schemas import chat as chat_schema
from oasst_inference_s... | inference/server/oasst_inference_server/routes/chats.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Enabling Threads and Retry for Web Chat | 2,232 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/chats.py
PR Title: Enabling Threads and Retry for Web Chat
BEFORE CODE: ucr: UserChatRepository
async with deps.manual_user_chat_repository(user_id) as ucr:
work_parameters = inference.... | 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 chat_schema.MessageTimeoutException as e:
logger.warning(f"Message timed out before work could be initiated: {e.message.id=}")
await handle_timeout(message=e.message)
finally:
_add_dequeue(pending_futu... | No issues found. | except chat_schema.MessageTimeoutException as e:
logger.warning(f"Message timed out before work could be initiated: {e.message.id=}")
await handle_timeout(message=e.message)
finally:
_add_dequeue(pending_futu... | @@ -327,7 +327,7 @@ async def handle_token_response(
deps.redis_client,
message_id=work_response_container.message_id,
)
- await message_queue.enqueue(response.json(), expire=settings.message_queue_expire)
+ await message_queue.enqueue(response.json())
work_response_container.num_respo... | inference/server/oasst_inference_server/routes/workers.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Computing message queue positions | 2,235 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/workers.py
PR Title: Computing message queue positions
BEFORE CODE: except chat_schema.MessageTimeoutException as e:
logger.warning(f"Message timed out before work could... | 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. | |
to run the galactica-125m model, using the webgpt dataset only (as opposed to all
the datasets listed in defaults in config.yaml) and treat each digit as a separate token.
""",
)
parser.add_argument("--local_rank", type=int, default=-1)
parser.add_argument("--deepspeed", action="store_tr... | No issues found. | to run the galactica-125m model, using the webgpt dataset only (as opposed to all
the datasets listed in defaults in config.yaml) and treat each digit as a separate token.
""",
)
parser.add_argument("--local_rank", type=int, default=-1)
parser.add_argument("--deepspeed", action="store_tr... | @@ -67,6 +67,7 @@ def compute_loss(self, model, inputs, return_outputs=False):
outputs = model(
input_ids=inputs["input_ids"],
attention_mask=inputs.get("attention_mask", None),
+ use_cache=False,
)
loss = self.loss_fct(outputs.get("logits"), targets, ma... | model/model_training/trainer_sft.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Fix GPTNeoX-20B training | 2,240 | LAION-AI/Open-Assistant | 29,137 | Python | dvruette | Review this pull request change.
Language: Python
File: model/model_training/trainer_sft.py
PR Title: Fix GPTNeoX-20B training
BEFORE CODE: to run the galactica-125m model, using the webgpt dataset only (as opposed to all
the datasets listed in defaults in config.yaml) and treat each digit as a sepa... | 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 { ChakraProvider, cookieStorageManagerSSR, localStorageManager } from "@chakra-ui/react";
import { PropsWithChildren } from "react";
import { theme } from "./Theme";
export function Chakra({ cookie, children }: PropsWithChildren<{ cookie: string }>) {
const colorModeManager = typeof cookie === "string" ? coo... | No issues found. | import { ChakraProvider, cookieStorageManagerSSR, localStorageManager } from "@chakra-ui/react";
import { PropsWithChildren } from "react";
import { theme } from "./Theme";
export function Chakra({ cookie, children }: PropsWithChildren<{ cookie: string }>) {
const colorModeManager = typeof cookie === "string" ? coo... | @@ -1,24 +1,14 @@
import { ChakraProvider, cookieStorageManagerSSR, localStorageManager } from "@chakra-ui/react";
+import { PropsWithChildren } from "react";
import { theme } from "./Theme";
-export function Chakra({ cookies, children }) {
- const colorModeManager = typeof cookies === "string" ? cookieStorageMa... | website/src/styles/Chakra.tsx | 0 | TypeScript | 1 | none | 16 | 15 | 15 | true | Expose env vars globally | 2,244 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/styles/Chakra.tsx
PR Title: Expose env vars globally
BEFORE CODE: import { ChakraProvider, cookieStorageManagerSSR, localStorageManager } from "@chakra-ui/react";
import { PropsWithChildren } from "react";
import { theme } from "./Theme";
exp... | 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. | |
) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
# self: transformers.models.llama.modeling_llama.LlamaAttention
if output_attentions:
warnings.warn("Output attentions is not supported for patched `LlamaAttention`, returning `None` instead.")
bsz, q_len, _ = hidden_st... | No issues found. | ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
# self: transformers.models.llama.modeling_llama.LlamaAttention
if output_attentions:
warnings.warn("Output attentions is not supported for patched `LlamaAttention`, returning `None` instead.")
bsz, q_len, _ = hidden_st... | @@ -0,0 +1,88 @@
+import warnings
+from typing import Optional, Tuple
+
+import torch
+import torch.nn as nn
+from transformers.models.llama.modeling_llama import LlamaAttention, apply_rotary_pos_emb
+
+from .patching_utils import compute_flash_attention
+
+
+def llama_forward_with_flash_attn(
+ self: LlamaAttention... | model/model_training/models/patching_llama.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Flash attention support for Llama | 2,277 | LAION-AI/Open-Assistant | 29,137 | Python | dvruette | Review this pull request change.
Language: Python
File: model/model_training/models/patching_llama.py
PR Title: Flash attention support for Llama
BEFORE CODE: ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
# self: transformers.models.llama.modeling_llama.LlamaAttention
if 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. | |
},
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 = token.name;
session.user.tosAcceptanceDate... | No issues found. | },
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 = token.name;
session.user.tosAcceptanceDate... | @@ -11,7 +11,6 @@ import EmailProvider from "next-auth/providers/email";
import { checkCaptcha } from "src/lib/captcha";
import { discordAvatarRefresh } from "src/lib/discord_avatar_refresh";
import { createApiClientFromUser } from "src/lib/oasst_client_factory";
-import { getInferenceTokens } from "src/lib/oasst_in... | website/src/pages/api/auth/[...nextauth].ts | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Use trusted clients for inference auth | 2,278 | 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: Use trusted clients for inference auth
BEFORE CODE: },
callbacks: {
/**
* Ensure we propagate the user's role when creating the session from the
* token.
*/
async session({ se... | 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. | |
load_sleep=${LOAD_SLEEP:-0}
mkdir -p $HF_HOME
echo -n "$HF_TOKEN" > $HF_HOME/token
export HUGGING_FACE_HUB_TOKEN=$HF_TOKEN
export MODEL_CONFIG_NAME=${MODEL_CONFIG_NAME:-"OA_SFT_Pythia_12B"}
export MODEL_ID=$(python get_model_config_prop.py model_id)
export QUANTIZE=$(python get_model_config_prop.py quantized)
echo ... | No issues found. | load_sleep=${LOAD_SLEEP:-0}
mkdir -p $HF_HOME
echo -n "$HF_TOKEN" > $HF_HOME/token
export HUGGING_FACE_HUB_TOKEN=$HF_TOKEN
export MODEL_CONFIG_NAME=${MODEL_CONFIG_NAME:-"OA_SFT_Pythia_12B"}
export MODEL_ID=$(python get_model_config_prop.py model_id)
export QUANTIZE=$(python get_model_config_prop.py quantized)
echo ... | @@ -1,14 +1,54 @@
#!/bin/bash
-mkdir -p $HOME/.cache/huggingface
-echo -n "$HF_TOKEN" > $HOME/.cache/huggingface/token
+export HF_HOME=${HF_HOME:-"$HOME/.cache/huggingface"}
+load_sleep=${LOAD_SLEEP:-0}
-python3 -m uvicorn --port 8883 basic_hf_server:app &
+mkdir -p $HF_HOME
+echo -n "$HF_TOKEN" > $HF_HOME/token
+... | inference/worker/worker_hf_main.sh | 0 | Shell | 1 | none | 16 | 51 | 51 | true | changed basic hf server to support quantization and streaming | 2,293 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: Shell
File: inference/worker/worker_hf_main.sh
PR Title: changed basic hf server to support quantization and streaming
BEFORE CODE: load_sleep=${LOAD_SLEEP:-0}
mkdir -p $HF_HOME
echo -n "$HF_TOKEN" > $HF_HOME/token
export HUGGING_FACE_HUB_TOKEN=$HF_TOKEN
export MODEL_CO... | 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 argument_parsing(notebook=False, notebook_args=None):
parser = argparse.ArgumentParser()
parser.add_argument(
"--datasets",
nargs="+",
required=True,
help="""
Multiple datasets can be passed to set different options.
For example, run as:
./check_d... | No issues found. |
def argument_parsing(notebook=False, notebook_args=None):
parser = argparse.ArgumentParser()
parser.add_argument(
"--datasets",
nargs="+",
required=True,
help="""
Multiple datasets can be passed to set different options.
For example, run as:
./check_d... | @@ -0,0 +1,125 @@
+import argparse
+from enum import Enum
+from pathlib import Path
+from typing import Any
+
+import pandas as pd
+import yaml
+from model_training.utils import _strtobool, get_dataset
+
+
+class Mode(Enum):
+ sft = "sft"
+ rm = "rm"
+ rl = "rl"
+
+ def config_name(self) -> str:
+ ma... | model/model_training/check_dataset_counts.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | add dataset counts script | 2,294 | LAION-AI/Open-Assistant | 29,137 | Python | CloseChoice | Review this pull request change.
Language: Python
File: model/model_training/check_dataset_counts.py
PR Title: add dataset counts script
BEFORE CODE:
def argument_parsing(notebook=False, notebook_args=None):
parser = argparse.ArgumentParser()
parser.add_argument(
"--datasets",
nargs="+",
... | 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 HellaSwagDataset(Dataset):
"""
Dataset class to use data from https://arxiv.org/pdf/1905.07830.pdf
for Reward modeling
"""
name = "hellaswag"
def __init__(self, split) -> None:
super().__init__()
np.random.seed(SEED)
self.dataset_list = []
dataset = loa... | No issues found. |
class HellaSwagDataset(Dataset):
"""
Dataset class to use data from https://arxiv.org/pdf/1905.07830.pdf
for Reward modeling
"""
name = "hellaswag"
def __init__(self, split) -> None:
super().__init__()
np.random.seed(SEED)
self.dataset_list = []
dataset = loa... | @@ -78,27 +78,53 @@ class HFDataset(Dataset):
name = "hfsummary"
- def __init__(self, split=None) -> None:
+ def __init__(self, split=None, subset="axis") -> None:
super().__init__()
+ # axis subset contains splits 'test' and 'validation'
+ # comparisons subset contains splits 'tra... | model/model_training/custom_datasets/rank_datasets.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | two more datasets | 2,301 | LAION-AI/Open-Assistant | 29,137 | Python | mikegarts | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/rank_datasets.py
PR Title: two more datasets
BEFORE CODE:
class HellaSwagDataset(Dataset):
"""
Dataset class to use data from https://arxiv.org/pdf/1905.07830.pdf
for Reward modeling
"""
name = "hella... | 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 len(self.dataset_list)
def __getitem__(self, idx):
context, completions = self.dataset_list[idx].values()
return [context], completions
class HFDataset(Dataset):
"""
Dataset class to use data from openai/summarize_from_feedback for Reward modeling.
Summaries ranked by o... | No issues found. | return len(self.dataset_list)
def __getitem__(self, idx):
context, completions = self.dataset_list[idx].values()
return [context], completions
class HFDataset(Dataset):
"""
Dataset class to use data from openai/summarize_from_feedback for Reward modeling.
Summaries ranked by o... | @@ -40,7 +40,6 @@ def __getitem__(self, idx):
class HellaSwagDataset(Dataset):
-
"""
Dataset class to use data from https://arxiv.org/pdf/1905.07830.pdf
for Reward modeling
@@ -70,7 +69,6 @@ def __getitem__(self, idx):
class HFDataset(Dataset):
-
"""
Dataset class to use data from ope... | model/model_training/custom_datasets/rank_datasets.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | revert instructor code and fix a bug in anthropic ds parsing | 2,307 | LAION-AI/Open-Assistant | 29,137 | Python | mikegarts | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/rank_datasets.py
PR Title: revert instructor code and fix a bug in anthropic ds parsing
BEFORE CODE: return len(self.dataset_list)
def __getitem__(self, idx):
context, completions = self.dataset_list[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. | |
# dialogue_collator.py line 36
)
gen_sequences = outputs.sequences[:, inputs["input_ids"].shape[-1] :]
for output in gen_sequences:
decoded = tokenizer.decode(
output, truncate_before_pattern=[r"\n\n^#", "^'''", "\n\n\n"], skip_special_tokens=True
... | No issues found. | # dialogue_collator.py line 36
)
gen_sequences = outputs.sequences[:, inputs["input_ids"].shape[-1] :]
for output in gen_sequences:
decoded = tokenizer.decode(
output, truncate_before_pattern=[r"\n\n^#", "^'''", "\n\n\n"], skip_special_tokens=True
... | @@ -0,0 +1,247 @@
+"""
+ Augment oasst dataset with sft generated results
+
+ You can use augment new response using a model with bad response, ie non SFT model
+
+ had to do this in a quick fashion, please tolerate the hackiness in the code
+
+"""
+import json
+import os
+
+# so far load_oasst_export is prett... | model/model_training/tools/augment_oasst.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | fix deepspeed issue on trainer_rm.py, add crossentropy support | 2,321 | LAION-AI/Open-Assistant | 29,137 | Python | theblackcat102 | Review this pull request change.
Language: Python
File: model/model_training/tools/augment_oasst.py
PR Title: fix deepspeed issue on trainer_rm.py, add crossentropy support
BEFORE CODE: # dialogue_collator.py line 36
)
gen_sequences = outputs.sequences[:, inputs["input_ids"].shape[-1] :]... | 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. | |
work_end_at: datetime.datetime | None = Field(None)
worker_id: str | None = Field(None, foreign_key="worker.id")
worker_compat_hash: str | None = Field(None, index=True)
worker_config: inference.WorkerConfig | None = Field(None, sa_column=sa.Column(pg.JSONB))
score: int = Field(0)
@property
... | No issues found. | work_end_at: datetime.datetime | None = Field(None)
worker_id: str | None = Field(None, foreign_key="worker.id")
worker_compat_hash: str | None = Field(None, index=True)
worker_config: inference.WorkerConfig | None = Field(None, sa_column=sa.Column(pg.JSONB))
score: int = Field(0)
@property
... | @@ -1,18 +1,18 @@
import datetime
-from uuid import uuid4
import sqlalchemy as sa
import sqlalchemy.dialects.postgresql as pg
from oasst_inference_server.schemas import chat as chat_schema
from oasst_shared.schemas import inference
from sqlmodel import Field, Relationship, SQLModel
+from uuid_extensions import ... | inference/server/oasst_inference_server/models/chat.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | inference: use uuid v7 for most of table | 2,327 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/models/chat.py
PR Title: inference: use uuid v7 for most of table
BEFORE CODE: work_end_at: datetime.datetime | None = Field(None)
worker_id: str | None = Field(None, foreign_key="worker.id")
worker_compat_ha... | 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. | |
parser.add_argument("--rng_seed", type=int, help="rng seed")
parser.add_argument("--show_dataset_stats", action="store_true", help="Show dataset stats", default=False)
parser.set_defaults(deepspeed=False)
if notebook:
args, remaining = parser.parse_known_args(notebook_args)
else:
ar... | No issues found. | parser.add_argument("--rng_seed", type=int, help="rng seed")
parser.add_argument("--show_dataset_stats", action="store_true", help="Show dataset stats", default=False)
parser.set_defaults(deepspeed=False)
if notebook:
args, remaining = parser.parse_known_args(notebook_args)
else:
ar... | @@ -1,9 +1,8 @@
import argparse
import logging
import os
-from typing import Callable, Dict, List, Literal, Optional, Tuple, Union
+from typing import Callable, Literal, Optional, Union
-import bitsandbytes
import datasets
import torch
from model_training.custom_datasets.ranking_collator import RankingDataColla... | model/model_training/trainer_rm.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Improve scores of small 1.4B reward model.. | 2,329 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/trainer_rm.py
PR Title: Improve scores of small 1.4B reward model..
BEFORE CODE: parser.add_argument("--rng_seed", type=int, help="rng seed")
parser.add_argument("--show_dataset_stats", action="store_true", help="Show dataset stats... | 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. | |
stats: Partial<{ [time in LeaderboardTimeFrame]: LeaderboardEntity }>;
title?: string;
}) => {
const { t } = useTranslation("leaderboard");
return (
<SurveyCard>
<Title>{title || t("your_stats")}</Title>
{[LeaderboardTimeFrame.day, LeaderboardTimeFrame.week, LeaderboardTimeFrame.month, Leaderbo... | No issues found. | stats: Partial<{ [time in LeaderboardTimeFrame]: LeaderboardEntity }>;
title?: string;
}) => {
const { t } = useTranslation("leaderboard");
return (
<SurveyCard>
<Title>{title || t("your_stats")}</Title>
{[LeaderboardTimeFrame.day, LeaderboardTimeFrame.week, LeaderboardTimeFrame.month, Leaderbo... | @@ -0,0 +1,74 @@
+import { Box, Flex, Grid, Text } from "@chakra-ui/react";
+import { useTranslation } from "next-i18next";
+import { ReactNode } from "react";
+import { getTypeSafei18nKey } from "src/lib/i18n";
+import { LeaderboardEntity, LeaderboardTimeFrame } from "src/types/Leaderboard";
+
+import { SurveyCard } f... | website/src/components/Account/UserStats.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | stats are shown on the admin | 2,330 | LAION-AI/Open-Assistant | 29,137 | Python | vivasvan1 | Review this pull request change.
Language: TypeScript
File: website/src/components/Account/UserStats.tsx
PR Title: stats are shown on the admin
BEFORE CODE: stats: Partial<{ [time in LeaderboardTimeFrame]: LeaderboardEntity }>;
title?: string;
}) => {
const { t } = useTranslation("leaderboard");
return (
... | 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 { BrowserEnv } from "src/lib/browserEnv";
import { SWRConfig, SWRConfiguration } from "swr";
import nextI18NextConfig from "../../next-i18next.config.js";
import { Chakra } from "../styles/Chakra";
type AppPropsWithLayout = AppProps & {
Component: NextPageWithLayout;
} & AppInitialProps;
const swrConfig: SW... | No issues found. | import { BrowserEnv } from "src/lib/browserEnv";
import { SWRConfig, SWRConfiguration } from "swr";
import nextI18NextConfig from "../../next-i18next.config.js";
import { Chakra } from "../styles/Chakra";
type AppPropsWithLayout = AppProps & {
Component: NextPageWithLayout;
} & AppInitialProps;
const swrConfig: SW... | @@ -10,33 +10,25 @@ import React, { useEffect } from "react";
import { FlagsProvider } from "react-feature-flags";
import { getDefaultLayout, NextPageWithLayout } from "src/components/Layout";
import flags from "src/flags";
+import { BrowserEnv } from "src/lib/browserEnv";
import { SWRConfig, SWRConfiguration } fro... | website/src/pages/_app.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Use SSR for every pages | 2,334 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/pages/_app.tsx
PR Title: Use SSR for every pages
BEFORE CODE: import { BrowserEnv } from "src/lib/browserEnv";
import { SWRConfig, SWRConfiguration } from "swr";
import nextI18NextConfig from "../../next-i18next.config.js";
import { Chakra } 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. | |
};
const Chat = () => {
const { t } = useTranslation(["common", "chat"]);
const router = useRouter();
const { data, mutate: refreshChatList } = useSWR<GetChatsResponse>("/api/chat", get, { revalidateOnFocus: true });
const { trigger: newChatTrigger } = useSWRMutation<{ id: string }>("/api/chat", post);
con... | No issues found. | };
const Chat = () => {
const { t } = useTranslation(["common", "chat"]);
const router = useRouter();
const { data, mutate: refreshChatList } = useSWR<GetChatsResponse>("/api/chat", get, { revalidateOnFocus: true });
const { trigger: newChatTrigger } = useSWRMutation<{ id: string }>("/api/chat", post);
con... | @@ -1,11 +1,12 @@
-import { Box, Button, Divider, Flex, Progress, Text } from "@chakra-ui/react";
+import { Box, Button, Divider, Flex, Grid, Progress, Text } from "@chakra-ui/react";
import { GetServerSideProps } from "next";
import Head from "next/head";
import Link from "next/link";
import { useRouter } from "ne... | website/src/pages/chat/index.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Add delete chat button | 2,345 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/pages/chat/index.tsx
PR Title: Add delete chat button
BEFORE CODE: };
const Chat = () => {
const { t } = useTranslation(["common", "chat"]);
const router = useRouter();
const { data, mutate: refreshChatList } = useSWR<GetChatsResponse>(... | 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 { useEffect } from "react";
import { UseFormReturn } from "react-hook-form";
import { ChatConfigForm } from "src/types/Chat";
const CHAT_CONFIG_KEY = "CHAT_CONFIG";
export const getCachedChatForm = (): ChatConfigForm | null => {
if (typeof localStorage !== "undefined") {
const oldConfig = localStorage.ge... | No issues found. | import { useEffect } from "react";
import { UseFormReturn } from "react-hook-form";
import { ChatConfigForm } from "src/types/Chat";
const CHAT_CONFIG_KEY = "CHAT_CONFIG";
export const getCachedChatForm = (): ChatConfigForm | null => {
if (typeof localStorage !== "undefined") {
const oldConfig = localStorage.ge... | @@ -0,0 +1,25 @@
+import { useEffect } from "react";
+import { UseFormReturn } from "react-hook-form";
+import { ChatConfigForm } from "src/types/Chat";
+
+const CHAT_CONFIG_KEY = "CHAT_CONFIG";
+
+export const getCachedChatForm = (): ChatConfigForm | null => {
+ if (typeof localStorage !== "undefined") {
+ const o... | website/src/hooks/chat/useCacheConfig.ts | 0 | TypeScript | 1 | none | 16 | 26 | 26 | true | Persist chat config | 2,367 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/hooks/chat/useCacheConfig.ts
PR Title: Persist chat config
BEFORE CODE: import { useEffect } from "react";
import { UseFormReturn } from "react-hook-form";
import { ChatConfigForm } from "src/types/Chat";
const CHAT_CONFIG_KEY = "CHAT_CONFIG";... | 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. | |
export { getServerSideProps } from "src/lib/defaultServerSideProps";
import { Box, Card, CardBody, Grid, Heading, Stack, Text, useColorModeValue } from "@chakra-ui/react";
import Head from "next/head";
import { useTranslation } from "next-i18next";
import React from "react";
import { TeamMember } from "src/components/T... | No issues found. | export { getServerSideProps } from "src/lib/defaultServerSideProps";
import { Box, Card, CardBody, Grid, Heading, Stack, Text, useColorModeValue } from "@chakra-ui/react";
import Head from "next/head";
import { useTranslation } from "next-i18next";
import React from "react";
import { TeamMember } from "src/components/T... | @@ -1,28 +1,14 @@
export { getServerSideProps } from "src/lib/defaultServerSideProps";
-import {
- Avatar,
- Badge,
- Box,
- Card,
- CardBody,
- Flex,
- Grid,
- Heading,
- Stack,
- Text,
- useColorModeValue,
-} from "@chakra-ui/react";
-import { Github } from "lucide-react";
+import { Box, Card, CardBody, G... | website/src/pages/team.tsx | 0 | TypeScript | 1 | none | 16 | 50 | 50 | true | Added stability, and added sponsors to inference interface | 2,383 | LAION-AI/Open-Assistant | 29,137 | Python | yk | Review this pull request change.
Language: TypeScript
File: website/src/pages/team.tsx
PR Title: Added stability, and added sponsors to inference interface
BEFORE CODE: export { getServerSideProps } from "src/lib/defaultServerSideProps";
import { Box, Card, CardBody, Grid, Heading, Stack, Text, useColorModeValue } ... | 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. | |
export const convertToBackendUserCore = <T extends { accounts: Account[]; id: string; name: string }>(
user: T
): BackendUserCore => {
// If there are no linked accounts, just use what we have locally.
if (user.accounts.length === 0) {
return {
id: user.id,
display_name: user.name,
auth_meth... | No issues found. | export const convertToBackendUserCore = <T extends { accounts: Account[]; id: string; name: string }>(
user: T
): BackendUserCore => {
// If there are no linked accounts, just use what we have locally.
if (user.accounts.length === 0) {
return {
id: user.id,
display_name: user.name,
auth_meth... | @@ -1,5 +1,6 @@
import { Account } from "@prisma/client";
import prisma from "src/lib/prismadb";
+import { AuthMethod } from "src/types/Providers";
import type { BackendUserCore } from "src/types/Users";
/**
@@ -39,7 +40,7 @@ export const convertToBackendUserCore = <T extends { accounts: Account[]; id: st
retu... | website/src/lib/users.ts | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Google auth website | 2,391 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/lib/users.ts
PR Title: Google auth website
BEFORE CODE: export const convertToBackendUserCore = <T extends { accounts: Account[]; id: string; name: string }>(
user: T
): BackendUserCore => {
// If there are no linked accounts, just use what... | 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 pydantic
class Settings(pydantic.BaseSettings):
safety_model_name: str = "shahules786/blade2blade-t5-base"
settings = Settings()
| No issues found. | import pydantic
class Settings(pydantic.BaseSettings):
safety_model_name: str = "shahules786/blade2blade-t5-base"
settings = Settings()
| @@ -0,0 +1,8 @@
+import pydantic
+
+
+class Settings(pydantic.BaseSettings):
+ safety_model_name: str = "shahules786/blade2blade-t5-base"
+
+
+settings = Settings() | inference/safety/settings.py | 0 | Python | 1 | none | 16 | 9 | 9 | true | Add safety server to inference | 2,449 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/safety/settings.py
PR Title: Add safety server to inference
BEFORE CODE: import pydantic
class Settings(pydantic.BaseSettings):
safety_model_name: str = "shahules786/blade2blade-t5-base"
settings = Settings()
AFTER CODE: import pydantic
c... | 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 deleted field to user
Revision ID: f0e18084aae4
Revises: 78f16015b904
Create Date: 2023-04-12 20:33:28.239793
"""
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision = "f0e18084aae4"
down_revision = "78f16015b904"
branch_labels = None
depends_on = None
def upgrad... | No issues found. | """Add deleted field to user
Revision ID: f0e18084aae4
Revises: 78f16015b904
Create Date: 2023-04-12 20:33:28.239793
"""
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision = "f0e18084aae4"
down_revision = "78f16015b904"
branch_labels = None
depends_on = None
def upgrad... | @@ -0,0 +1,27 @@
+"""Add deleted field to user
+
+Revision ID: f0e18084aae4
+Revises: 78f16015b904
+Create Date: 2023-04-12 20:33:28.239793
+
+"""
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "f0e18084aae4"
+down_revision = "78f16015b904"
+branch_labels = None... | inference/server/alembic/versions/2023_04_12_2033-f0e18084aae4_add_deleted_field_to_user.py | 0 | Python | 1 | none | 16 | 28 | 28 | true | Add support for user deletion | 2,486 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/server/alembic/versions/2023_04_12_2033-f0e18084aae4_add_deleted_field_to_user.py
PR Title: Add support for user deletion
BEFORE CODE: """Add deleted field to user
Revision ID: f0e18084aae4
Revises: 78f16015b904
Create Date: 2023-04-12 20:33:28.2397... | 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. | |
export const ChatListBase = memo(function ChatListBase({
isSideBar,
chats,
...props
}: CardProps & { isSideBar: boolean; chats?: GetChatsResponse }) {
const { data: response } = useSWR<GetChatsResponse>(chats ? null : API_ROUTES.LIST_CHAT, get, {
fallbackData: chats,
});
const { t } = useTranslation(["c... | No issues found. | export const ChatListBase = memo(function ChatListBase({
isSideBar,
chats,
...props
}: CardProps & { isSideBar: boolean; chats?: GetChatsResponse }) {
const { data: response } = useSWR<GetChatsResponse>(chats ? null : API_ROUTES.LIST_CHAT, get, {
fallbackData: chats,
});
const { t } = useTranslation(["c... | @@ -1,10 +1,9 @@
import "simplebar-react/dist/simplebar.min.css";
-import { Button, Card, CardProps } from "@chakra-ui/react";
+import { Card, CardProps } from "@chakra-ui/react";
import { Plus } from "lucide-react";
-import Link from "next/link";
import { useTranslation } from "next-i18next";
-import { memo } fro... | website/src/components/Chat/ChatListBase.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Don't refresh the page after each message | 2,491 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatListBase.tsx
PR Title: Don't refresh the page after each message
BEFORE CODE: export const ChatListBase = memo(function ChatListBase({
isSideBar,
chats,
...props
}: CardProps & { isSideBar: boolean; chats?: GetChatsRes... | 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 safe_content to message
Revision ID: ea19bbc743f9
Revises: 401eef162771
Create Date: 2023-04-14 22:37:41.373382
"""
import sqlalchemy as sa
import sqlmodel
from alembic import op
# revision identifiers, used by Alembic.
revision = "ea19bbc743f9"
down_revision = "401eef162771"
branch_labels = None
depends_on =... | No issues found. | """Add safe_content to message
Revision ID: ea19bbc743f9
Revises: 401eef162771
Create Date: 2023-04-14 22:37:41.373382
"""
import sqlalchemy as sa
import sqlmodel
from alembic import op
# revision identifiers, used by Alembic.
revision = "ea19bbc743f9"
down_revision = "401eef162771"
branch_labels = None
depends_on =... | @@ -0,0 +1,34 @@
+"""Add safe_content to message
+
+Revision ID: ea19bbc743f9
+Revises: 401eef162771
+Create Date: 2023-04-14 22:37:41.373382
+
+"""
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "ea19bbc743f9"
+down_revision = "401eef162771"
+b... | inference/server/alembic/versions/2023_04_29_1739-ea19bbc743f9_add_safe_content_to_message.py | 0 | Python | 1 | none | 16 | 35 | 35 | true | Inference worker inform backend on safety intervention | 2,505 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/server/alembic/versions/2023_04_29_1739-ea19bbc743f9_add_safe_content_to_message.py
PR Title: Inference worker inform backend on safety intervention
BEFORE CODE: """Add safe_content to message
Revision ID: ea19bbc743f9
Revises: 401eef162771
Create D... | 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 { withoutRole } from "src/lib/auth";
import { isChatEnable } from "src/lib/isChatEnable";
import { createInferenceClient } from "src/lib/oasst_inference_client";
const handler = withoutRole("banned", async (req, res, token) => {
if (!isChatEnable()) {
return res.status(404).end();
}
const client = cre... | No issues found. | import { withoutRole } from "src/lib/auth";
import { isChatEnable } from "src/lib/isChatEnable";
import { createInferenceClient } from "src/lib/oasst_inference_client";
const handler = withoutRole("banned", async (req, res, token) => {
if (!isChatEnable()) {
return res.status(404).end();
}
const client = cre... | @@ -0,0 +1,17 @@
+import { withoutRole } from "src/lib/auth";
+import { isChatEnable } from "src/lib/isChatEnable";
+import { createInferenceClient } from "src/lib/oasst_inference_client";
+
+const handler = withoutRole("banned", async (req, res, token) => {
+ if (!isChatEnable()) {
+ return res.status(404).end();
... | website/src/pages/api/chat/title.ts | 0 | TypeScript | 1 | none | 16 | 18 | 18 | true | implement update chat title UI | 2,507 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/pages/api/chat/title.ts
PR Title: implement update chat title UI
BEFORE CODE: import { withoutRole } from "src/lib/auth";
import { isChatEnable } from "src/lib/isChatEnable";
import { createInferenceClient } from "src/lib/oasst_inference_client... | 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 { ChatConfigFormData } from "src/types/Chat";
const CHAT_CONFIG_KEY = "CHAT_CONFIG";
export const setConfigCache = (config: ChatConfigFormData) => {
if (typeof localStorage !== "undefined") {
localStorage.setItem(CHAT_CONFIG_KEY, JSON.stringify(config));
}
};
export const getConfigCache = () => {
if... | No issues found. | import { ChatConfigFormData } from "src/types/Chat";
const CHAT_CONFIG_KEY = "CHAT_CONFIG";
export const setConfigCache = (config: ChatConfigFormData) => {
if (typeof localStorage !== "undefined") {
localStorage.setItem(CHAT_CONFIG_KEY, JSON.stringify(config));
}
};
export const getConfigCache = () => {
if... | @@ -0,0 +1,20 @@
+import { ChatConfigFormData } from "src/types/Chat";
+
+const CHAT_CONFIG_KEY = "CHAT_CONFIG";
+
+export const setConfigCache = (config: ChatConfigFormData) => {
+ if (typeof localStorage !== "undefined") {
+ localStorage.setItem(CHAT_CONFIG_KEY, JSON.stringify(config));
+ }
+};
+
+export const g... | website/src/utils/chat.ts | 0 | TypeScript | 1 | none | 16 | 21 | 21 | true | fix custom preset selection | 2,511 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/utils/chat.ts
PR Title: fix custom preset selection
BEFORE CODE: import { ChatConfigFormData } from "src/types/Chat";
const CHAT_CONFIG_KEY = "CHAT_CONFIG";
export const setConfigCache = (config: ChatConfigFormData) => {
if (typeof localSto... | 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 hidden field to chats
Revision ID: b66fd8f9da1f
Revises: f0e18084aae4
Create Date: 2023-04-14 16:11:35.361507
"""
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision = "b66fd8f9da1f"
down_revision = "f0e18084aae4"
branch_labels = None
depends_on = None
def upgrad... | No issues found. | """Add hidden field to chats
Revision ID: b66fd8f9da1f
Revises: f0e18084aae4
Create Date: 2023-04-14 16:11:35.361507
"""
import sqlalchemy as sa
from alembic import op
# revision identifiers, used by Alembic.
revision = "b66fd8f9da1f"
down_revision = "f0e18084aae4"
branch_labels = None
depends_on = None
def upgrad... | @@ -0,0 +1,27 @@
+"""Add hidden field to chats
+
+Revision ID: b66fd8f9da1f
+Revises: f0e18084aae4
+Create Date: 2023-04-14 16:11:35.361507
+
+"""
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "b66fd8f9da1f"
+down_revision = "f0e18084aae4"
+branch_labels = None... | inference/server/alembic/versions/2023_04_14_1611-b66fd8f9da1f_add_hidden_field_to_chats.py | 0 | Python | 1 | none | 16 | 28 | 28 | true | Add ability to hide (soft delete) chats to inference backend | 2,512 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/server/alembic/versions/2023_04_14_1611-b66fd8f9da1f_add_hidden_field_to_chats.py
PR Title: Add ability to hide (soft delete) chats to inference backend
BEFORE CODE: """Add hidden field to chats
Revision ID: b66fd8f9da1f
Revises: f0e18084aae4
Create... | 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. | |
formatted = ds_entry.get_formatted(Mode.sft, "<|endofline|>")
assert len(formatted) == 3
assert "length: 100" in formatted[0]
assert "quality: 1.0" in formatted[0]
assert "humor: 0.0" in formatted[0]
assert "creativity: 0.0" in formatted[0]
assert "Some context" in formatted[0]
assert f"... | No issues found. | formatted = ds_entry.get_formatted(Mode.sft, "<|endofline|>")
assert len(formatted) == 3
assert "length: 100" in formatted[0]
assert "quality: 1.0" in formatted[0]
assert "humor: 0.0" in formatted[0]
assert "creativity: 0.0" in formatted[0]
assert "Some context" in formatted[0]
assert f"... | @@ -0,0 +1,79 @@
+import pytest
+from model_training.custom_datasets.entities import Language, Mode
+from model_training.custom_datasets.formatting import QA_SPECIAL_TOKENS, DatasetEntry
+
+
+def test_dataset_entry_formatting_missing_lang():
+ ds_entry = DatasetEntry(
+ questions=["What is the capital of Fran... | model/model_training/tests/test_formatting.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Feature/dataset entry | 2,520 | LAION-AI/Open-Assistant | 29,137 | Python | CloseChoice | Review this pull request change.
Language: Python
File: model/model_training/tests/test_formatting.py
PR Title: Feature/dataset entry
BEFORE CODE: formatted = ds_entry.get_formatted(Mode.sft, "<|endofline|>")
assert len(formatted) == 3
assert "length: 100" in formatted[0]
assert "quality: 1.0" in fo... | 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 { data: session } = useSession();
const router = useRouter();
const { trigger } = useSWRMutation("/api/account/delete", del);
const executeDelete = useCallback(async () => {
try {
await trigger();
signOut({ callbackUrl: "/" });
} catch (err) {
console.error(err);
}
}, [tr... | No issues found. | const { data: session } = useSession();
const router = useRouter();
const { trigger } = useSWRMutation("/api/account/delete", del);
const executeDelete = useCallback(async () => {
try {
await trigger();
signOut({ callbackUrl: "/" });
} catch (err) {
console.error(err);
}
}, [tr... | @@ -0,0 +1,77 @@
+import { Button, Divider, Flex, Grid, Icon, ListItem, Text, UnorderedList } from "@chakra-ui/react";
+import Head from "next/head";
+import Link from "next/link";
+import React, { useCallback } from "react";
+export { getServerSideProps } from "src/lib/defaultServerSideProps";
+import { useRouter } fr... | website/src/pages/account/delete.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Add delete account page | 2,523 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/pages/account/delete.tsx
PR Title: Add delete account page
BEFORE CODE: const { data: session } = useSession();
const router = useRouter();
const { trigger } = useSWRMutation("/api/account/delete", del);
const executeDelete = useCallb... | 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. | |
"shp",
"hellaswag",
"webgpt",
]
def train_val_dataset(dataset, val_split=0.2) -> tuple[Dataset, Dataset | None]:
if val_split == 0:
return dataset, None
train_idx, val_idx = train_test_split(
list(range(len(dataset))), test_size=val_split, random_state=666, shuffle=True
)
... | No issues found. | "shp",
"hellaswag",
"webgpt",
]
def train_val_dataset(dataset, val_split=0.2) -> tuple[Dataset, Dataset | None]:
if val_split == 0:
return dataset, None
train_idx, val_idx = train_test_split(
list(range(len(dataset))), test_size=val_split, random_state=666, shuffle=True
)
... | @@ -10,6 +10,7 @@
from model_training.custom_datasets.prompt_dialogue import Gpt4All, load_oig_file
from model_training.custom_datasets.qa_datasets import (
SODA,
+ DatabricksDolly15k,
JokeExplaination,
QADataset,
SODADialogue,
@@ -159,6 +160,8 @@ def get_one_dataset(
train, eval = loa... | model/model_training/custom_datasets/__init__.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | remove references from dolly | 2,535 | LAION-AI/Open-Assistant | 29,137 | Python | CloseChoice | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/__init__.py
PR Title: remove references from dolly
BEFORE CODE: "shp",
"hellaswag",
"webgpt",
]
def train_val_dataset(dataset, val_split=0.2) -> tuple[Dataset, Dataset | None]:
if val_split == 0:
r... | 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. | |
raise fastapi.HTTPException(status_code=400, detail="Parent message is not an assistant message")
if msg_dict[parent_id].state != inference.MessageState.complete:
raise fastapi.HTTPException(status_code=400, detail="Parent message is not complete")
message = models.D... | No issues found. | raise fastapi.HTTPException(status_code=400, detail="Parent message is not an assistant message")
if msg_dict[parent_id].state != inference.MessageState.complete:
raise fastapi.HTTPException(status_code=400, detail="Parent message is not complete")
message = models.D... | @@ -228,16 +228,21 @@ async def add_report(self, message_id: str, reason: str, report_type: inference.
await self.session.refresh(report)
return report
- async def update_title(self, chat_id: str, title: str) -> None:
- logger.info(f"Updating title of chat {chat_id=}: {title=}")
+ async... | inference/server/oasst_inference_server/user_chat_repository.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Unify title update and visibility update inference endpoints | 2,627 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/user_chat_repository.py
PR Title: Unify title update and visibility update inference endpoints
BEFORE CODE: raise fastapi.HTTPException(status_code=400, detail="Parent message is not an assistant 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. | |
TextLabel.hate_speech,
TextLabel.sexual_content,
]
mandatory_labels_initial_prompt: Optional[list[TextLabel]] = [TextLabel.spam]
"""Mandatory labels in text-labeling tasks for initial prompts."""
mandatory_labels_assistant_reply: Optional[list[TextLabel]] = [TextLabel.spam]
"""Mand... | No issues found. | TextLabel.hate_speech,
TextLabel.sexual_content,
]
mandatory_labels_initial_prompt: Optional[list[TextLabel]] = [TextLabel.spam]
"""Mandatory labels in text-labeling tasks for initial prompts."""
mandatory_labels_assistant_reply: Optional[list[TextLabel]] = [TextLabel.spam]
"""Mand... | @@ -274,6 +274,9 @@ def validate_user_stats_intervals(cls, v: int):
TASK_VALIDITY_MINUTES: int = 60 * 24 * 2 # tasks expire after 2 days
+ DISCORD_API_KEY: str | None = None
+ DISCORD_CHANNEL_ID: str | None = None
+
class Config:
env_file = ".env"
env_file_encoding = "utf-8" | backend/oasst_backend/config.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | send flagged message notification to discord | 2,641 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: Python
File: backend/oasst_backend/config.py
PR Title: send flagged message notification to discord
BEFORE CODE: TextLabel.hate_speech,
TextLabel.sexual_content,
]
mandatory_labels_initial_prompt: Optional[list[TextLabel]] = [TextLabel.spam]
""... | 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. | |
}>(`/api/v1/messages/${message_id}/tree/state`);
}
/**
* Delete a message by its id
*/
async delete_message(message_id: string): Promise<void> {
return this.delete<void>(`/api/v1/messages/${message_id}`);
}
/**
* Undelete a message by its id
*/
async undelete_message(message_id: string... | No issues found. | }>(`/api/v1/messages/${message_id}/tree/state`);
}
/**
* Delete a message by its id
*/
async delete_message(message_id: string): Promise<void> {
return this.delete<void>(`/api/v1/messages/${message_id}`);
}
/**
* Undelete a message by its id
*/
async undelete_message(message_id: string... | @@ -400,7 +400,7 @@ export class OasstApiClient {
async set_tos_acceptance(user: BackendUserCore) {
// NOTE: we do a post here to force create the user if it does not exist
- const backendUser = await this.post<BackendUser>(`/api/v1/frontend_users/`, user);
+ const backendUser = await this.upsert_fronte... | website/src/lib/oasst_api_client.ts | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | create backend user when generate jwt | 2,702 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/lib/oasst_api_client.ts
PR Title: create backend user when generate jwt
BEFORE CODE: }>(`/api/v1/messages/${message_id}/tree/state`);
}
/**
* Delete a message by its id
*/
async delete_message(message_id: string): Promise<void>... | 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. | |
TEXT_DATASETS = {
"gutenberg_english": "sedthh/gutenberg_english", # Gutenberg eBooks in English
"gutenberg_multilang": "sedthh/gutenberg_multilang", # Gutenberg eBooks in foreign languages
"tv_dialogue": "sedthh/tv_dialogue", # TV and Movie dialogues and transcripts
"fd_dialogue": "sedthh/fd_dialogu... | No issues found. | TEXT_DATASETS = {
"gutenberg_english": "sedthh/gutenberg_english", # Gutenberg eBooks in English
"gutenberg_multilang": "sedthh/gutenberg_multilang", # Gutenberg eBooks in foreign languages
"tv_dialogue": "sedthh/tv_dialogue", # TV and Movie dialogues and transcripts
"fd_dialogue": "sedthh/fd_dialogu... | @@ -22,6 +22,7 @@
"oa_leet10k": "ehartford/oa_leet10k",
"LogicInference_OA": "KK04/LogicInference_OA",
"oa_dolly_15k": "OllieStanley/oa_dolly_15k",
+ "poetry_instruction": "checkai/poetry-instruction",
}
SAFETY_DATASETS = { | data/datasets/__init__.py | 0 | Python | 1 | none | 16 | 34 | 34 | true | add poetry dataset setup | 2,730 | LAION-AI/Open-Assistant | 29,137 | Python | check3D | Review this pull request change.
Language: Python
File: data/datasets/__init__.py
PR Title: add poetry dataset setup
BEFORE CODE: TEXT_DATASETS = {
"gutenberg_english": "sedthh/gutenberg_english", # Gutenberg eBooks in English
"gutenberg_multilang": "sedthh/gutenberg_multilang", # Gutenberg eBooks in fore... | 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 parent_id is None:
if len(chat.messages) > 0:
raise fastapi.HTTPException(status_code=400, detail="Trying to add first message to non-empty chat")
chat.title = content
else:
msg_dict = chat.get_msg_dict()
if parent_id not in msg_dict:
... | No issues found. | if parent_id is None:
if len(chat.messages) > 0:
raise fastapi.HTTPException(status_code=400, detail="Trying to add first message to non-empty chat")
chat.title = content
else:
msg_dict = chat.get_msg_dict()
if parent_id not in msg_dict:
... | @@ -14,12 +14,30 @@ class UserChatRepository(pydantic.BaseModel):
class Config:
arbitrary_types_allowed = True
- async def get_chats(self, include_hidden: bool = False) -> list[models.DbChat]:
+ async def get_chats(
+ self,
+ include_hidden: bool = False,
+ limit: int | None =... | inference/server/oasst_inference_server/user_chat_repository.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | support pagination for chat list | 2,773 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/user_chat_repository.py
PR Title: support pagination for chat list
BEFORE CODE: if parent_id is None:
if len(chat.messages) > 0:
raise fastapi.HTTPException(status_code=400, detail="Tr... | 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. | |
@validator("length")
def above_zero(cls, v) -> int:
if v is not None and v < 0:
raise ValueError(f"Length cannot be lower than 0. Received {v}")
return v
@validator("quality", "humor", "creativity")
def between_0_1(cls, v, field: ModelField) -> float:
if v is not No... | No issues found. |
@validator("length")
def above_zero(cls, v) -> int:
if v is not None and v < 0:
raise ValueError(f"Length cannot be lower than 0. Received {v}")
return v
@validator("quality", "humor", "creativity")
def between_0_1(cls, v, field: ModelField) -> float:
if v is not No... | @@ -1,7 +1,8 @@
from itertools import zip_longest
from random import shuffle
-from model_training.custom_datasets.entities import Language, Mode
+from langcodes import Language
+from model_training.custom_datasets.entities import Mode
from pydantic import BaseModel, validator
from pydantic.fields import ModelFiel... | model/model_training/custom_datasets/formatting.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | update entities | 2,809 | LAION-AI/Open-Assistant | 29,137 | Python | CloseChoice | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/formatting.py
PR Title: update entities
BEFORE CODE:
@validator("length")
def above_zero(cls, v) -> int:
if v is not None and v < 0:
raise ValueError(f"Length cannot be lower than 0. Received {v... | 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 (model.includes("Llama")) {
const OA = await createInferenceClient(user.username, user.id);
if (!chatId) {
let chat = await OA.create_chat();
chatId = chat.id;
}
let prompter_message = await OA.post_prompter_message({
chat_id: chatId,
content: message,
parent_id: paren... | No issues found. | if (model.includes("Llama")) {
const OA = await createInferenceClient(user.username, user.id);
if (!chatId) {
let chat = await OA.create_chat();
chatId = chat.id;
}
let prompter_message = await OA.post_prompter_message({
chat_id: chatId,
content: message,
parent_id: paren... | @@ -0,0 +1,75 @@
+import { createInferenceClient } from "../modules/inference/client.js";
+import { HfInference } from "@huggingface/inference";
+const hf = new HfInference(process.env.HUGGINGFACE_TOKEN);
+
+export default async function chat(
+ model,
+ user,
+ message,
+ chatId?,
+ parentId?,
+ presets?,
+ pre... | discord-bots/oa-bot-js/src/modules/chat.ts | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Discord bot: Pythia model | 2,831 | LAION-AI/Open-Assistant | 29,137 | Python | mrloldev | Review this pull request change.
Language: TypeScript
File: discord-bots/oa-bot-js/src/modules/chat.ts
PR Title: Discord bot: Pythia model
BEFORE CODE: if (model.includes("Llama")) {
const OA = await createInferenceClient(user.username, user.id);
if (!chatId) {
let chat = await OA.create_chat();
... | 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. | |
token = self.tokens.pop()
end_tokens.append(token)
end_sequence = token.text + end_sequence
if end_sequence in self.stop_sequences:
break
else:
self.tokens.extend(reversed(end_tokens))
yield from ... | No issues found. | token = self.tokens.pop()
end_tokens.append(token)
end_sequence = token.text + end_sequence
if end_sequence in self.stop_sequences:
break
else:
self.tokens.extend(reversed(end_tokens))
yield from ... | @@ -103,9 +103,8 @@ def text_to_events(text: str, seed: int | None = None, pause: float = 0.0):
def lorem_events(seed):
- sentence = lorem.sentence()
- time.sleep(1)
- yield from text_to_events(sentence, seed=seed, pause=0.5)
+ sentence = lorem.paragraph()
+ yield from text_to_events(sentence, seed=... | inference/worker/utils.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Implement auto scroll when stream message | 2,839 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: Python
File: inference/worker/utils.py
PR Title: Implement auto scroll when stream message
BEFORE CODE: token = self.tokens.pop()
end_tokens.append(token)
end_sequence = token.text + end_sequence
if end_sequen... | 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. | |
(e: ChangeEvent<HTMLSelectElement>) => {
const newPresetName = e.target.value;
const config =
newPresetName === customPresetName
? customPresetDefaultValue
: presets.find((preset) => preset.name === newPresetName)!.sampling_parameters;
for (const [key, value] of Object... | No issues found. | (e: ChangeEvent<HTMLSelectElement>) => {
const newPresetName = e.target.value;
const config =
newPresetName === customPresetName
? customPresetDefaultValue
: presets.find((preset) => preset.name === newPresetName)!.sampling_parameters;
for (const [key, value] of Object... | @@ -17,9 +17,9 @@ import {
Tooltip,
} from "@chakra-ui/react";
import { useTranslation } from "next-i18next";
-import { ChangeEvent, memo, useCallback, useState } from "react";
-import { Controller, useFormContext, useWatch } from "react-hook-form";
-import { ChatConfigFormData, SamplingParameters } from "src/type... | website/src/components/Chat/ChatConfigForm.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Fix saving model between chats | 2,843 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConfigForm.tsx
PR Title: Fix saving model between chats
BEFORE CODE: (e: ChangeEvent<HTMLSelectElement>) => {
const newPresetName = e.target.value;
const config =
newPresetName === customPresetName
... | 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. | |
chats: chatResponse.chats.map((chat) => {
if (chat.id === chatId) {
return {
...chat,
title,
};
}
return chat;
}),
})) || []),
],
false
);
},
[mut... | No issues found. | chats: chatResponse.chats.map((chat) => {
if (chat.id === chatId) {
return {
...chat,
title,
};
}
return chat;
}),
})) || []),
],
false
);
},
[mut... | @@ -5,51 +5,67 @@ import { Plus } from "lucide-react";
import { useTranslation } from "next-i18next";
import { memo, useCallback } from "react";
import SimpleBar from "simplebar-react";
-import { get } from "src/lib/api";
-import { API_ROUTES } from "src/lib/routes";
import { GetChatsResponse } from "src/types/Chat... | website/src/components/Chat/ChatListBase.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | feat(website): support pagination for chat list | 2,860 | LAION-AI/Open-Assistant | 29,137 | Python | mtosity | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatListBase.tsx
PR Title: feat(website): support pagination for chat list
BEFORE CODE: chats: chatResponse.chats.map((chat) => {
if (chat.id === chatId) {
return {
...... | 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. | |
{
...columnHelper.accessor("display_name", {
header: "Name",
}),
filterable: true,
meta: {
cellProps: (x) => {
return { style: { overflow: "hidden" } };
},
},
},
columnHelper.accessor("role", {
header: "Role",
}),
columnHelper.accessor((user) => user.user_id, ... | No issues found. | {
...columnHelper.accessor("display_name", {
header: "Name",
}),
filterable: true,
meta: {
cellProps: (x) => {
return { style: { overflow: "hidden" } };
},
},
},
columnHelper.accessor("role", {
header: "Role",
}),
columnHelper.accessor((user) => user.user_id, ... | @@ -1,5 +1,5 @@
import { Card, CardBody, IconButton } from "@chakra-ui/react";
-import { createColumnHelper } from "@tanstack/react-table";
+import { createColumnHelper, Cell } from "@tanstack/react-table";
import { Pencil } from "lucide-react";
import Link from "next/link";
import { memo, useState } from "react";
... | website/src/components/UserTable.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Fix display name spill-over (#2851) | 2,863 | LAION-AI/Open-Assistant | 29,137 | Python | scorbett123 | Review this pull request change.
Language: TypeScript
File: website/src/components/UserTable.tsx
PR Title: Fix display name spill-over (#2851)
BEFORE CODE: {
...columnHelper.accessor("display_name", {
header: "Name",
}),
filterable: true,
meta: {
cellProps: (x) => {
return { st... | 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. | |
isSending: boolean;
onSubmit: () => void;
queueInfo: QueueInfo | null;
};
// eslint-disable-next-line react/display-name
export const ChatForm = forwardRef<HTMLTextAreaElement, ChatFormProps>((props, ref) => {
const { isSending, onSubmit: onSubmit, queueInfo } = props;
const { t } = useTranslation("chat");
... | No issues found. | isSending: boolean;
onSubmit: () => void;
queueInfo: QueueInfo | null;
};
// eslint-disable-next-line react/display-name
export const ChatForm = forwardRef<HTMLTextAreaElement, ChatFormProps>((props, ref) => {
const { isSending, onSubmit: onSubmit, queueInfo } = props;
const { t } = useTranslation("chat");
... | @@ -35,12 +35,7 @@ export const ChatForm = forwardRef<HTMLTextAreaElement, ChatFormProps>((props, r
[onSubmit]
);
return (
- <Box
- as="form"
- maxWidth={{ base: "3xl", "2xl": "4xl" }}
- onSubmit={handleSubmit}
- className="pt-2 pb-4 w-full mx-auto"
- >
+ <Box as="form" maxWidth=... | website/src/components/Chat/ChatForm.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | update chat warning design | 2,924 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatForm.tsx
PR Title: update chat warning design
BEFORE CODE: isSending: boolean;
onSubmit: () => void;
queueInfo: QueueInfo | null;
};
// eslint-disable-next-line react/display-name
export const ChatForm = forwardRef<HT... | 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 read_message_trees(input_file_path: str | Path) -> Iterable[ExportMessageTree]:
for x in read_oasst_jsonl(input_file_path):
assert isinstance(x, ExportMessageTree)
yield x
def read_message_tree_list(
input_file_path: str | Path,
filter: Optional[Callable[[ExportMessageTree], bool]] = N... | No issues found. | def read_message_trees(input_file_path: str | Path) -> Iterable[ExportMessageTree]:
for x in read_oasst_jsonl(input_file_path):
assert isinstance(x, ExportMessageTree)
yield x
def read_message_tree_list(
input_file_path: str | Path,
filter: Optional[Callable[[ExportMessageTree], bool]] = N... | @@ -4,6 +4,7 @@
from typing import Callable, Iterable, Optional, TextIO
import pydantic
+from datasets import load_dataset
from .schemas import ExportMessageNode, ExportMessageTree
@@ -17,22 +18,24 @@ def open_jsonl_read(input_file_path: str | Path) -> TextIO:
return input_file_path.open("r", encoding... | oasst-data/oasst_data/reader.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Add data loader for HF oasst1 | 2,951 | LAION-AI/Open-Assistant | 29,137 | Python | grgau | Review this pull request change.
Language: Python
File: oasst-data/oasst_data/reader.py
PR Title: Add data loader for HF oasst1
BEFORE CODE: def read_message_trees(input_file_path: str | Path) -> Iterable[ExportMessageTree]:
for x in read_oasst_jsonl(input_file_path):
assert isinstance(x, ExportMessageT... | 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 { AxiosError } from "axios";
import { withoutRole } from "src/lib/auth";
import { isChatEnable } from "src/lib/isChatEnable";
import { createInferenceClient } from "src/lib/oasst_inference_client";
import { InferencePostAssistantMessageParams } from "src/types/Chat";
const handler = withoutRole("banned", async ... | No issues found. | import { AxiosError } from "axios";
import { withoutRole } from "src/lib/auth";
import { isChatEnable } from "src/lib/isChatEnable";
import { createInferenceClient } from "src/lib/oasst_inference_client";
import { InferencePostAssistantMessageParams } from "src/types/Chat";
const handler = withoutRole("banned", async ... | @@ -1,3 +1,4 @@
+import { AxiosError } from "axios";
import { withoutRole } from "src/lib/auth";
import { isChatEnable } from "src/lib/isChatEnable";
import { createInferenceClient } from "src/lib/oasst_inference_client";
@@ -9,12 +10,16 @@ const handler = withoutRole("banned", async (req, res, token) => {
}
c... | website/src/pages/api/chat/assistant_message.ts | 0 | TypeScript | 1 | none | 16 | 26 | 26 | true | Handle chat error | 2,959 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/pages/api/chat/assistant_message.ts
PR Title: Handle chat error
BEFORE CODE: import { AxiosError } from "axios";
import { withoutRole } from "src/lib/auth";
import { isChatEnable } from "src/lib/isChatEnable";
import { createInferenceClient } 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. | |
// eslint-disable-next-line react/display-name
export const ChatForm = forwardRef<HTMLTextAreaElement, ChatFormProps>((props, forwardedRef) => {
const { isSending, onSubmit: onSubmit, queueInfo } = props;
const { t } = useTranslation("chat");
const handleSubmit = useCallback(
(e: SyntheticEvent) => {
e.... | No issues found. | // eslint-disable-next-line react/display-name
export const ChatForm = forwardRef<HTMLTextAreaElement, ChatFormProps>((props, forwardedRef) => {
const { isSending, onSubmit: onSubmit, queueInfo } = props;
const { t } = useTranslation("chat");
const handleSubmit = useCallback(
(e: SyntheticEvent) => {
e.... | @@ -1,8 +1,9 @@
-import { Box, CircularProgress, Flex, Textarea } from "@chakra-ui/react";
+import { Box, CircularProgress, Flex, Textarea, useBreakpointValue } from "@chakra-ui/react";
import { Send } from "lucide-react";
import { useTranslation } from "next-i18next";
-import { forwardRef, KeyboardEvent, SyntheticEv... | website/src/components/Chat/ChatForm.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Auto focus chat input | 2,998 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatForm.tsx
PR Title: Auto focus chat input
BEFORE CODE: // eslint-disable-next-line react/display-name
export const ChatForm = forwardRef<HTMLTextAreaElement, ChatFormProps>((props, forwardedRef) => {
const { isSending, onSu... | 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. | |
: presets.find((preset) => preset.name === newPresetName)!.sampling_parameters;
for (const [key, value] of Object.entries(config) as Array<[keyof SamplingParameters, number]>) {
setValue(key, value, { shouldDirty: true }); // force dirty so the ChatConfigSaver will update the cache
}
... | No issues found. | : presets.find((preset) => preset.name === newPresetName)!.sampling_parameters;
for (const [key, value] of Object.entries(config) as Array<[keyof SamplingParameters, number]>) {
setValue(key, value, { shouldDirty: true }); // force dirty so the ChatConfigSaver will update the cache
}
... | @@ -92,7 +92,7 @@ export const ChatConfigForm = memo(function ChatConfigForm() {
const { control, getValues, register, setValue } = useFormContext<ChatConfigFormData>();
const selectedModel = getValues("model_config_name"); // have to use getValues to here to access latest value
- const plugins = getValues("pl... | website/src/components/Chat/ChatConfigForm.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Fixed PluginChooser button width problem | 3,014 | LAION-AI/Open-Assistant | 29,137 | Python | draganjovanovich | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConfigForm.tsx
PR Title: Fixed PluginChooser button width problem
BEFORE CODE: : presets.find((preset) => preset.name === newPresetName)!.sampling_parameters;
for (const [key, value] of Object.entries(config... | 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. | |
safety_rots=safety_rots,
),
)
logger.debug(f"Safe prompt: {prompt}")
stream_response = None
token_buffer = utils.TokenBuffer(stop_sequences=parameters.stop)
if model_config.is_lorem:
stream_events = utils.lorem_events(parameters.seed)
... | No issues found. | safety_rots=safety_rots,
),
)
logger.debug(f"Safe prompt: {prompt}")
stream_response = None
token_buffer = utils.TokenBuffer(stop_sequences=parameters.stop)
if model_config.is_lorem:
stream_events = utils.lorem_events(parameters.seed)
... | @@ -1,58 +1,25 @@
import re
-import threading
from concurrent import futures
import chat_chain
import interface
import requests
-import sseclient
import transformers
import utils
import websocket
-from chat_chain_prompts import ASSISTANT_PREFIX, END_SEQ, OBSERVATION_SEQ, START_SEQ, THOUGHT_SEQ
+from chat_chai... | inference/worker/work.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Fix plugins | 3,017 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/worker/work.py
PR Title: Fix plugins
BEFORE CODE: safety_rots=safety_rots,
),
)
logger.debug(f"Safe prompt: {prompt}")
stream_response = None
token_buffer = utils.TokenBuffer(stop_... | 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. | |
pe="50px"
borderRadius="lg"
maxLength={100}
autoFocus
></Input>
<Flex
position="absolute"
alignContent="center"
style={{
insetInlineEnd: `8px`,
}}
gap="1.5"
zIndex={1... | No issues found. | pe="50px"
borderRadius="lg"
maxLength={100}
autoFocus
></Input>
<Flex
position="absolute"
alignContent="center"
style={{
insetInlineEnd: `8px`,
}}
gap="1.5"
zIndex={1... | @@ -1,10 +1,32 @@
-import { Box, Button, CircularProgress, Flex, Input, Tooltip, useBoolean, useOutsideClick } from "@chakra-ui/react";
-import { Check, EyeOff, LucideIcon, Pencil, X } from "lucide-react";
+import {
+ AlertDialog,
+ AlertDialogBody,
+ AlertDialogContent,
+ AlertDialogFooter,
+ AlertDialogHeader,
+... | website/src/components/Chat/ChatListItem.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | add delete function | 3,019 | LAION-AI/Open-Assistant | 29,137 | Python | ruoniw | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatListItem.tsx
PR Title: add delete function
BEFORE CODE: pe="50px"
borderRadius="lg"
maxLength={100}
autoFocus
></Input>
<Flex
position="absolute... | 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. | |
handler: setIsEditing.off,
});
const handleKeydown = useCallback(
(e: KeyboardEvent) => {
if (e.key === "Escape") {
setIsEditing.off();
}
if (e.key === "Enter" && !e.shiftKey) {
handleEditSubmit();
}
},
[handleEditSubmit, setIsEditing]
);
const { onCopy,... | No issues found. | handler: setIsEditing.off,
});
const handleKeydown = useCallback(
(e: KeyboardEvent) => {
if (e.key === "Escape") {
setIsEditing.off();
}
if (e.key === "Enter" && !e.shiftKey) {
handleEditSubmit();
}
},
[handleEditSubmit, setIsEditing]
);
const { onCopy,... | @@ -22,6 +22,7 @@ import { BaseMessageEntry } from "../Messages/BaseMessageEntry";
import { BaseMessageEmojiButton } from "../Messages/MessageEmojiButton";
import { MessageInlineEmojiRow } from "../Messages/MessageInlineEmojiRow";
import { WorkParametersDisplay } from "./WorkParameters";
+import { EncourageMessage }... | website/src/components/Chat/ChatMessageEntry.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Created feedback encouragement message | 3,057 | LAION-AI/Open-Assistant | 29,137 | Python | amitag2 | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatMessageEntry.tsx
PR Title: Created feedback encouragement message
BEFORE CODE: handler: setIsEditing.off,
});
const handleKeydown = useCallback(
(e: KeyboardEvent) => {
if (e.key === "Escape") {
se... | 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. | |
name="k50-Plugins",
description="Top-k sampling with k=50 and temperature=0.35",
sampling_parameters=inference.SamplingParameters(
max_new_tokens=1024,
temperature=0.35,
top_k=50,
repetition_penalty=(1 / 0.90),
),
),
ParameterConfig... | No issues found. | name="k50-Plugins",
description="Top-k sampling with k=50 and temperature=0.35",
sampling_parameters=inference.SamplingParameters(
max_new_tokens=1024,
temperature=0.35,
top_k=50,
repetition_penalty=(1 / 0.90),
),
),
ParameterConfig... | @@ -1,4 +1,5 @@
import asyncio
+import json
import aiohttp
import fastapi
@@ -121,8 +122,14 @@ async def fetch_plugin(url: str, retries: int = 3, timeout: float = 5.0) -> infe
content_type = response.headers.get("Content-Type")
if response.status == 200:
- ... | inference/server/oasst_inference_server/routes/configs.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Plugins fixes | 3,070 | LAION-AI/Open-Assistant | 29,137 | Python | draganjovanovich | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/routes/configs.py
PR Title: Plugins fixes
BEFORE CODE: name="k50-Plugins",
description="Top-k sampling with k=50 and temperature=0.35",
sampling_parameters=inference.SamplingParameters(
... | 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 { useEffect, useMemo, useRef } from "react";
import { get } from "src/lib/api";
import { API_ROUTES } from "src/lib/routes";
import { GetChatsResponse } from "src/types/Chat";
import useSWRInfinite from "swr/infinite";
export type ChatListViewSelection = "visible" | "visible_hidden";
export function useListCha... | No issues found. | import { useEffect, useMemo, useRef } from "react";
import { get } from "src/lib/api";
import { API_ROUTES } from "src/lib/routes";
import { GetChatsResponse } from "src/types/Chat";
import useSWRInfinite from "swr/infinite";
export type ChatListViewSelection = "visible" | "visible_hidden";
export function useListCha... | @@ -1,25 +1,28 @@
import { useEffect, useMemo, useRef } from "react";
import { get } from "src/lib/api";
+import { API_ROUTES } from "src/lib/routes";
import { GetChatsResponse } from "src/types/Chat";
import useSWRInfinite from "swr/infinite";
-export function useListChatPagination(initialChats?: GetChatsRespons... | website/src/components/Chat/useListChatPagination.ts | 0 | TypeScript | 1 | none | 16 | 47 | 47 | true | Show hidden chats in chat list | 3,075 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/useListChatPagination.ts
PR Title: Show hidden chats in chat list
BEFORE CODE: import { useEffect, useMemo, useRef } from "react";
import { get } from "src/lib/api";
import { API_ROUTES } from "src/lib/routes";
import { GetChats... | 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. | |
<ChatParameterField
{...item}
value={getValues(name)} // need to call getValues here, react-hook-form not trigger rerender when call setValue manually
onChange={onChange}
name={name}
isDisabled={selectedPresetName ... | No issues found. | <ChatParameterField
{...item}
value={getValues(name)} // need to call getValues here, react-hook-form not trigger rerender when call setValue manually
onChange={onChange}
name={name}
isDisabled={selectedPresetName ... | @@ -19,14 +19,18 @@ import {
import { useTranslation } from "next-i18next";
import { ChangeEvent, memo, useCallback, useEffect, useRef, useState } from "react";
import { Controller, useFormContext, UseFormSetValue } from "react-hook-form";
+import SimpleBar from "simplebar-react";
import { ChatConfigFormData, Model... | website/src/components/Chat/ChatConfigForm.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Allow to save custom preset | 3,079 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatConfigForm.tsx
PR Title: Allow to save custom preset
BEFORE CODE: <ChatParameterField
{...item}
value={getValues(name)} // need to call getValues here, react-hook-form not ... | 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 text search
Revision ID: 1b6e3ae16e9d
Revises: 9db92d504f64
Create Date: 2023-05-07 21:29:35.545612
"""
import sqlalchemy as sa
from alembic import op
from sqlalchemy.dialects import postgresql
# revision identifiers, used by Alembic.
revision = "1b6e3ae16e9d"
down_revision = "9db92d504f64"
branch_labels = Non... | No issues found. | """add text search
Revision ID: 1b6e3ae16e9d
Revises: 9db92d504f64
Create Date: 2023-05-07 21:29:35.545612
"""
import sqlalchemy as sa
from alembic import op
from sqlalchemy.dialects import postgresql
# revision identifiers, used by Alembic.
revision = "1b6e3ae16e9d"
down_revision = "9db92d504f64"
branch_labels = Non... | @@ -0,0 +1,25 @@
+"""add text search
+
+Revision ID: 1b6e3ae16e9d
+Revises: 9db92d504f64
+Create Date: 2023-05-07 21:29:35.545612
+"""
+import sqlalchemy as sa
+from alembic import op
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision = "1b6e3ae16e9d"
+down_revision = "9db... | backend/alembic/versions/2023_05_07_2129-1b6e3ae16e9d_add_text_search.py | 0 | Python | 1 | none | 16 | 26 | 26 | true | Backend support for message search | 3,080 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: backend/alembic/versions/2023_05_07_2129-1b6e3ae16e9d_add_text_search.py
PR Title: Backend support for message search
BEFORE CODE: """add text search
Revision ID: 1b6e3ae16e9d
Revises: 9db92d504f64
Create Date: 2023-05-07 21:29:35.545612
"""
import sqlalchemy... | 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. | |
from oasst_backend.config import settings
from oasst_backend.models.message import Message
ROOT_ENDPOINT = "https://discord.com/api/v10"
@celery_app.task(name="send_new_report_message")
def send_new_report_message(message: Message, label_text: str, user_id: UUID):
"""
Send a message to the Discord channel wh... | No issues found. | from oasst_backend.config import settings
from oasst_backend.models.message import Message
ROOT_ENDPOINT = "https://discord.com/api/v10"
@celery_app.task(name="send_new_report_message")
def send_new_report_message(message: Message, label_text: str, user_id: UUID):
"""
Send a message to the Discord channel wh... | @@ -2,13 +2,24 @@
import requests
from loguru import logger
+from oasst_backend.celery_worker import app as celery_app
from oasst_backend.config import settings
from oasst_backend.models.message import Message
ROOT_ENDPOINT = "https://discord.com/api/v10"
-def send_new_report_message(message: Message, labe... | backend/oasst_backend/utils/discord.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Make send report message from discord a celery task. | 3,089 | LAION-AI/Open-Assistant | 29,137 | Python | espoirMur | Review this pull request change.
Language: Python
File: backend/oasst_backend/utils/discord.py
PR Title: Make send report message from discord a celery task.
BEFORE CODE: from oasst_backend.config import settings
from oasst_backend.models.message import Message
ROOT_ENDPOINT = "https://discord.com/api/v10"
@cele... | 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. | |
# format to QnA
def qna_wrapper():
def create_qna(row):
# make a random number
random_num = random.randint(0, 2)
# extract rows' vals
con_type = row["Type"]
lang = row["Language"]
word1 = row["Word1"]
word2 = row["Word2"]
score_percent = row["Score"]... | No issues found. | # format to QnA
def qna_wrapper():
def create_qna(row):
# make a random number
random_num = random.randint(0, 2)
# extract rows' vals
con_type = row["Type"]
lang = row["Language"]
word1 = row["Word1"]
word2 = row["Word2"]
score_percent = row["Score"]... | @@ -0,0 +1,72 @@
+import json
+import random
+from dataclasses import dataclass
+
+import pandas as pd
+import random_stuff
+from datasets import load_dataset
+
+random.seed(42)
+
+
+# format to QnA
+def qna_wrapper():
+ def create_qna(row):
+ # make a random number
+ random_num = random.randint(0, 2)
... | data/datasets/semantics_ws_qna_oa/data_process.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Add semantics WS dataset | 3,200 | LAION-AI/Open-Assistant | 29,137 | Python | echo0x22 | Review this pull request change.
Language: Python
File: data/datasets/semantics_ws_qna_oa/data_process.py
PR Title: Add semantics WS dataset
BEFORE CODE: # format to QnA
def qna_wrapper():
def create_qna(row):
# make a random number
random_num = random.randint(0, 2)
# extract rows' vals... | 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. | |
tool_names = None
final_input = (
f"{input_prompt}{eos_token}{V2_ASST_PREFIX}\n{prev_chain_response}{OBSERVATION_SEQ} {tool_response}"
)
inner_prompt = prepare_prompt(
final_input,
prompt_template,
memor... | No issues found. | tool_names = None
final_input = (
f"{input_prompt}{eos_token}{V2_ASST_PREFIX}\n{prev_chain_response}{OBSERVATION_SEQ} {tool_response}"
)
inner_prompt = prepare_prompt(
final_input,
prompt_template,
memor... | @@ -25,9 +25,6 @@
from oasst_shared.schemas import inference
from settings import settings
-# Max depth of retries for tool usage
-MAX_DEPTH = 6
-
# Exclude tools description from final prompt. Saves ctx space but can hurt output
# quality especially if truncation kicks in. Dependent on model used
REMOVE_TOOLS_F... | inference/worker/chat_chain.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Add inference server setting to control max depth of plugins | 3,227 | LAION-AI/Open-Assistant | 29,137 | Python | olliestanley | Review this pull request change.
Language: Python
File: inference/worker/chat_chain.py
PR Title: Add inference server setting to control max depth of plugins
BEFORE CODE: tool_names = None
final_input = (
f"{input_prompt}{eos_token}{V2_ASST_PREFIX}\n{prev_chain_response}... | 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. | |
message: inference.MessageRead | None = None
class MessageResponseEvent(pydantic.BaseModel):
event_type: Literal["message"] = "message"
message: inference.MessageRead
class SafePromptResponseEvent(pydantic.BaseModel):
event_type: Literal["safe_prompt"] = "safe_prompt"
safe_prompt: str
messag... | No issues found. | message: inference.MessageRead | None = None
class MessageResponseEvent(pydantic.BaseModel):
event_type: Literal["message"] = "message"
message: inference.MessageRead
class SafePromptResponseEvent(pydantic.BaseModel):
event_type: Literal["safe_prompt"] = "safe_prompt"
safe_prompt: str
messag... | @@ -46,8 +46,23 @@ class SafePromptResponseEvent(pydantic.BaseModel):
message: inference.MessageRead
+class PluginIntermediateResponseEvent(pydantic.BaseModel):
+ event_type: Literal["plugin_intermediate"] = "plugin_intermediate"
+ current_plugin_thought: str
+ current_plugin_action_taken: str
+ cu... | inference/server/oasst_inference_server/schemas/chat.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Plugins intermediate steps reporting | 3,238 | LAION-AI/Open-Assistant | 29,137 | Python | draganjovanovich | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/schemas/chat.py
PR Title: Plugins intermediate steps reporting
BEFORE CODE: message: inference.MessageRead | None = None
class MessageResponseEvent(pydantic.BaseModel):
event_type: Literal["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. | |
app = FastAPI()
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
@app.get("/roast-user/", operation_id="roastUser")
async def roast_user(
roasting_style: str = Query(..., description="style of roasting and malice tha... | No issues found. |
app = FastAPI()
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
@app.get("/roast-user/", operation_id="roastUser")
async def roast_user(
roasting_style: str = Query(..., description="style of roasting and malice tha... | @@ -0,0 +1,66 @@
+import json
+from pathlib import Path
+
+from fastapi import FastAPI, Query, Request, Response
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.openapi.utils import get_openapi
+
+folder_path = Path(__file__).parent
+
+app = FastAPI()
+
+app.add_middleware(
+ CORSMiddleware,
+ a... | inference/server/oasst_inference_server/plugins/gale_roaster/main.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Add Dragan's web retriever plugin | 3,277 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | Review this pull request change.
Language: Python
File: inference/server/oasst_inference_server/plugins/gale_roaster/main.py
PR Title: Add Dragan's web retriever plugin
BEFORE CODE:
app = FastAPI()
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"]... | 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_message_revisions
Revision ID: c181661eba3a
Revises: 1b6e3ae16e9d
Create Date: 2023-06-06 15:05:58.079120
"""
import sqlalchemy as sa
from alembic import op
from sqlalchemy.dialects import postgresql
# revision identifiers, used by Alembic.
revision = "c181661eba3a"
down_revision = "1b6e3ae16e9d"
branch_label... | No issues found. | """add_message_revisions
Revision ID: c181661eba3a
Revises: 1b6e3ae16e9d
Create Date: 2023-06-06 15:05:58.079120
"""
import sqlalchemy as sa
from alembic import op
from sqlalchemy.dialects import postgresql
# revision identifiers, used by Alembic.
revision = "c181661eba3a"
down_revision = "1b6e3ae16e9d"
branch_label... | @@ -0,0 +1,50 @@
+"""add_message_revisions
+
+Revision ID: c181661eba3a
+Revises: 1b6e3ae16e9d
+Create Date: 2023-06-06 15:05:58.079120
+
+"""
+import sqlalchemy as sa
+from alembic import op
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision = "c181661eba3a"
+down_revisio... | backend/alembic/versions/2023_06_06_1505-c181661eba3a_add_message_revisions.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Moderator Message Editing | 3,280 | LAION-AI/Open-Assistant | 29,137 | Python | someone13574 | Review this pull request change.
Language: Python
File: backend/alembic/versions/2023_06_06_1505-c181661eba3a_add_message_revisions.py
PR Title: Moderator Message Editing
BEFORE CODE: """add_message_revisions
Revision ID: c181661eba3a
Revises: 1b6e3ae16e9d
Create Date: 2023-06-06 15:05:58.079120
"""
import sqlalc... | 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. | |
onReport={showReportPopup}
message={message}
/>
<LabelMessagePopup message={message} show={labelPopupOpen} onClose={closeLabelPopup} />
<ReportPopup messageId={message.id} show={reportPopupOpen} onClose={closeReportPopup} />
</MessageInlineEmojiR... | No issues found. | onReport={showReportPopup}
message={message}
/>
<LabelMessagePopup message={message} show={labelPopupOpen} onClose={closeLabelPopup} />
<ReportPopup messageId={message.id} show={reportPopupOpen} onClose={closeReportPopup} />
</MessageInlineEmojiR... | @@ -9,7 +9,6 @@ import {
MenuList,
Portal,
SimpleGrid,
- Text,
Tooltip,
useColorModeValue,
useDisclosure,
@@ -39,7 +38,6 @@ import { LabelMessagePopup } from "src/components/Messages/LabelPopup";
import { MessageEmojiButton } from "src/components/Messages/MessageEmojiButton";
import { ReportPopup }... | website/src/components/Messages/MessageTableEntry.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Edit history viewer | 3,322 | LAION-AI/Open-Assistant | 29,137 | Python | someone13574 | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/MessageTableEntry.tsx
PR Title: Edit history viewer
BEFORE CODE: onReport={showReportPopup}
message={message}
/>
<LabelMessagePopup message={message} show={labelPopupOpen} ... | 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. | |
borderInlineStartColor: "gray.300",
_dark: {
borderInlineStartColor: "gray.500",
},
},
"table tbody tr": {
borderBottomColor: "gray.400",
_dark: {
borderBottomColor: "gray.700",
},
},
"table thead tr": {
borderBottomColor: "gray.400",
borderBottomWidth: "1px",
_da... | No issues found. | borderInlineStartColor: "gray.300",
_dark: {
borderInlineStartColor: "gray.500",
},
},
"table tbody tr": {
borderBottomColor: "gray.400",
_dark: {
borderBottomColor: "gray.700",
},
},
"table thead tr": {
borderBottomColor: "gray.400",
borderBottomWidth: "1px",
_da... | @@ -1,3 +1,5 @@
+import "katex/dist/katex.min.css";
+
import {
Box,
Button,
@@ -17,13 +19,11 @@ import { useTranslation } from "next-i18next";
import { memo, MouseEvent, useMemo, useState } from "react";
import ReactMarkdown from "react-markdown";
import type { ReactMarkdownOptions } from "react-markdown/lib/... | website/src/components/Messages/RenderedMarkdown.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Bug Fix: Delete feature (Fixes three issues) | 3,383 | LAION-AI/Open-Assistant | 29,137 | Python | CertifiedJoon | Review this pull request change.
Language: TypeScript
File: website/src/components/Messages/RenderedMarkdown.tsx
PR Title: Bug Fix: Delete feature (Fixes three issues)
BEFORE CODE: borderInlineStartColor: "gray.300",
_dark: {
borderInlineStartColor: "gray.500",
},
},
"table tbody tr": {
bo... | 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. | |
})) || []),
],
false
);
},
[mutateChatResponses]
);
const removeItemFromList = useCallback(
({ chatId }: { chatId: string }) => {
mutateChatResponses(
(chatResponses) => [
...(chatResponses?.map((chatResponse) => ({
...chatResponse,
... | No issues found. | })) || []),
],
false
);
},
[mutateChatResponses]
);
const removeItemFromList = useCallback(
({ chatId }: { chatId: string }) => {
mutateChatResponses(
(chatResponses) => [
...(chatResponses?.map((chatResponse) => ({
...chatResponse,
... | @@ -14,9 +14,9 @@ import { ChatListViewSelection, useListChatPagination } from "./useListChatPagin
export const ChatListBase = memo(function ChatListBase({
allowViews,
- minHeight,
+ noScrollbar,
...props
-}: CardProps & { allowViews?: boolean; minHeight?: string }) {
+}: CardProps & { allowViews?: boolean; ... | website/src/components/Chat/ChatListBase.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | fix chat list double scrollbar | 3,385 | LAION-AI/Open-Assistant | 29,137 | Python | notmd | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/ChatListBase.tsx
PR Title: fix chat list double scrollbar
BEFORE CODE: })) || []),
],
false
);
},
[mutateChatResponses]
);
const removeItemFromList = useCallback(
({ chatId }: { c... | 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. | |
onReview: () => void;
onSubmit: () => void;
onSkip: () => void;
}
export const TaskControls = ({
task,
taskStatus,
isLoading,
isRejecting,
isSubmitting,
onEdit,
onReview,
onSubmit,
onSkip,
}: TaskControlsProps) => {
const { t } = useTranslation();
return (
<Card>
{isLoading && <P... | No issues found. | onReview: () => void;
onSubmit: () => void;
onSkip: () => void;
}
export const TaskControls = ({
task,
taskStatus,
isLoading,
isRejecting,
isSubmitting,
onEdit,
onReview,
onSubmit,
onSkip,
}: TaskControlsProps) => {
const { t } = useTranslation();
return (
<Card>
{isLoading && <P... | @@ -1,11 +1,13 @@
-import { Button, Card, Flex, IconButton, Progress, Tooltip } from "@chakra-ui/react";
+import { Card, Flex, IconButton, Progress, Tooltip } from "@chakra-ui/react";
import { Edit2 } from "lucide-react";
import { useTranslation } from "next-i18next";
import { SubmitButton } from "src/components/But... | website/src/components/Survey/TaskControls.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Added confirmation dialog when skipping after entering text | 3,479 | LAION-AI/Open-Assistant | 29,137 | Python | MehrJonas | Review this pull request change.
Language: TypeScript
File: website/src/components/Survey/TaskControls.tsx
PR Title: Added confirmation dialog when skipping after entering text
BEFORE CODE: onReview: () => void;
onSubmit: () => void;
onSkip: () => void;
}
export const TaskControls = ({
task,
taskStatus,
... | 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 { data: session, status } = useSession();
const isAdminOrMod = useHasAnyRole(["admin", "moderator"]);
if (!session || status !== "authenticated") {
return null;
}
const options: MenuOption[] = [
{
name: t("dashboard"),
href: "/dashboard",
icon: Layout,
isExternal: false,
... | No issues found. | const { data: session, status } = useSession();
const isAdminOrMod = useHasAnyRole(["admin", "moderator"]);
if (!session || status !== "authenticated") {
return null;
}
const options: MenuOption[] = [
{
name: t("dashboard"),
href: "/dashboard",
icon: Layout,
isExternal: false,
... | @@ -72,7 +72,7 @@ export function UserMenu() {
<Menu>
<MenuButton border="solid" borderRadius="full" borderWidth="thin" borderColor={borderColor}>
<Box display="flex" alignItems="center" gap="3" p="1">
- <Avatar size="sm" src={session.user.image!} />
+ <Avatar size="sm" src={sessi... | website/src/components/Header/UserMenu.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Update profile image on login | 3,491 | LAION-AI/Open-Assistant | 29,137 | Python | AbdBarho | Review this pull request change.
Language: TypeScript
File: website/src/components/Header/UserMenu.tsx
PR Title: Update profile image on login
BEFORE CODE: const { data: session, status } = useSession();
const isAdminOrMod = useHasAnyRole(["admin", "moderator"]);
if (!session || status !== "authenticated") {
... | 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 { LucideProps } from "lucide-react";
export const MarkdownOffIcon = (props: LucideProps) => {
return (
<svg
xmlns="http://www.w3.org/2000/svg"
className="icon icon-tabler icon-tabler-markdown-off"
width={24}
height={24}
viewBox="0 0 24 24"
strokeWidth="2"
stroke="... | No issues found. | import { LucideProps } from "lucide-react";
export const MarkdownOffIcon = (props: LucideProps) => {
return (
<svg
xmlns="http://www.w3.org/2000/svg"
className="icon icon-tabler icon-tabler-markdown-off"
width={24}
height={24}
viewBox="0 0 24 24"
strokeWidth="2"
stroke="... | @@ -0,0 +1,26 @@
+import { LucideProps } from "lucide-react";
+
+export const MarkdownOffIcon = (props: LucideProps) => {
+ return (
+ <svg
+ xmlns="http://www.w3.org/2000/svg"
+ className="icon icon-tabler icon-tabler-markdown-off"
+ width={24}
+ height={24}
+ viewBox="0 0 24 24"
+ st... | website/src/components/icons/MarkdownOff.tsx | 0 | TypeScript | 1 | none | 16 | 27 | 27 | true | Add: option to view plaintext chat message. Issue/2508 | 3,496 | LAION-AI/Open-Assistant | 29,137 | Python | CertifiedJoon | Review this pull request change.
Language: TypeScript
File: website/src/components/icons/MarkdownOff.tsx
PR Title: Add: option to view plaintext chat message. Issue/2508
BEFORE CODE: import { LucideProps } from "lucide-react";
export const MarkdownOffIcon = (props: LucideProps) => {
return (
<svg
xmlns... | 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 { getValidDisplayName } from "src/lib/display_name_validation";
import prisma from "src/lib/prismadb";
import { getBatchFrontendUserIdFromBackendUser } from "src/lib/users";
import { AuthMethod } from "src/types/Providers";
export const updateUsersDisplayNames = <T extends { display_name: string; username: stri... | No issues found. | import { getValidDisplayName } from "src/lib/display_name_validation";
import prisma from "src/lib/prismadb";
import { getBatchFrontendUserIdFromBackendUser } from "src/lib/users";
import { AuthMethod } from "src/types/Providers";
export const updateUsersDisplayNames = <T extends { display_name: string; username: stri... | @@ -11,16 +11,16 @@ export const updateUsersDisplayNames = <T extends { display_name: string; userna
};
export const updateUsersProfilePictures = async <T extends { auth_method: AuthMethod; username: string }>(
- entires: T[]
+ entries: T[]
) => {
- const frontendUserIds = await getBatchFrontendUserIdFromBacken... | website/src/lib/leaderboard_utilities.ts | 0 | TypeScript | 1 | none | 16 | 30 | 30 | true | fix typos | 3,543 | LAION-AI/Open-Assistant | 29,137 | Python | RainRat | Review this pull request change.
Language: TypeScript
File: website/src/lib/leaderboard_utilities.ts
PR Title: fix typos
BEFORE CODE: import { getValidDisplayName } from "src/lib/display_name_validation";
import prisma from "src/lib/prismadb";
import { getBatchFrontendUserIdFromBackendUser } from "src/lib/users";
i... | 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 ModuleNotFoundError:
warnings.warn(
"""\nmodule flash_attn not found - either install:
pip3 install flash_attn
or run with:
--use_flash_attention=false """
)
exit(1)
if (resid_pdrop is None or resid_pdrop == 0.0) and not flash_attention:
... | No issues found. | except ModuleNotFoundError:
warnings.warn(
"""\nmodule flash_attn not found - either install:
pip3 install flash_attn
or run with:
--use_flash_attention=false """
)
exit(1)
if (resid_pdrop is None or resid_pdrop == 0.0) and not flash_attention:
... | @@ -6,12 +6,13 @@
import torch.nn as nn
import transformers
-from transformers import GPTNeoXForCausalLM, GPTNeoXModel, LlamaForCausalLM, LlamaModel
+from transformers import AutoConfig, GPTNeoXForCausalLM, GPTNeoXModel, LlamaForCausalLM, LlamaModel
from trlx.models.modeling_ppo import AutoModelForCausalLMWithHydr... | model/model_training/models/patching.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Add RoPE Interpolation | 3,564 | LAION-AI/Open-Assistant | 29,137 | Python | shahules786 | Review this pull request change.
Language: Python
File: model/model_training/models/patching.py
PR Title: Add RoPE Interpolation
BEFORE CODE: except ModuleNotFoundError:
warnings.warn(
"""\nmodule flash_attn not found - either install:
pip3 install flash_attn
or run with:
--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 torch
from model_training.models.patching import patch_model
from transformers import AutoTokenizer
from transformers.models.falcon.modeling_falcon import FalconForCausalLM
def test_flash_attention_falcon_patch(device="cuda:0"):
model_name = "tiiuae/falcon-7b"
dtype = torch.bfloat16
tokenizer = Au... | No issues found. | import torch
from model_training.models.patching import patch_model
from transformers import AutoTokenizer
from transformers.models.falcon.modeling_falcon import FalconForCausalLM
def test_flash_attention_falcon_patch(device="cuda:0"):
model_name = "tiiuae/falcon-7b"
dtype = torch.bfloat16
tokenizer = Au... | @@ -0,0 +1,34 @@
+import torch
+from model_training.models.patching import patch_model
+from transformers import AutoTokenizer
+from transformers.models.falcon.modeling_falcon import FalconForCausalLM
+
+
+def test_flash_attention_falcon_patch(device="cuda:0"):
+ model_name = "tiiuae/falcon-7b"
+ dtype = torch.bf... | model/model_training/tests/test_patched_falcon.py | 0 | Python | 1 | none | 16 | 35 | 35 | true | Add flash-attention patch for falcon-7b | 3,580 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/tests/test_patched_falcon.py
PR Title: Add flash-attention patch for falcon-7b
BEFORE CODE: import torch
from model_training.models.patching import patch_model
from transformers import AutoTokenizer
from transformers.models.falcon.modeling... | 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 handleOpen = () => {
setCustomInstructions(savedCustomInstructions);
setIsOpen(true);
};
const handleClose = () => setIsOpen(false);
const handleSave = () => {
setIsOpen(false);
onChange(customInstructions);
};
function createChangeHandler(key: string) {
return function handleInp... | No issues found. |
const handleOpen = () => {
setCustomInstructions(savedCustomInstructions);
setIsOpen(true);
};
const handleClose = () => setIsOpen(false);
const handleSave = () => {
setIsOpen(false);
onChange(customInstructions);
};
function createChangeHandler(key: string) {
return function handleInp... | @@ -0,0 +1,106 @@
+import React, { useState } from "react";
+import {
+ Button,
+ Modal,
+ ModalOverlay,
+ ModalContent,
+ ModalHeader,
+ ModalBody,
+ ModalFooter,
+ FormControl,
+ FormLabel,
+ Input,
+ Text,
+ Textarea,
+} from "@chakra-ui/react";
+import { BookOpen } from "lucide-react";
+import { useTran... | website/src/components/Chat/CustomInstructions.tsx | 0 | TypeScript | 1 | none | 16 | 51 | 51 | true | Custom instructions feature | 3,597 | LAION-AI/Open-Assistant | 29,137 | Python | draganjovanovich | Review this pull request change.
Language: TypeScript
File: website/src/components/Chat/CustomInstructions.tsx
PR Title: Custom instructions feature
BEFORE CODE:
const handleOpen = () => {
setCustomInstructions(savedCustomInstructions);
setIsOpen(true);
};
const handleClose = () => setIsOpen(false);
... | 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 train_val_dataset(dataset, val_split=0.2) -> tuple[Dataset, Dataset | None]:
if val_split == 0:
return dataset, None
train_idx, val_idx = train_test_split(
list(range(len(dataset))), test_size=val_split, random_state=666, shuffle=True
)
return Subset(dataset, train_idx), Subset(da... | No issues found. |
def train_val_dataset(dataset, val_split=0.2) -> tuple[Dataset, Dataset | None]:
if val_split == 0:
return dataset, None
train_idx, val_idx = train_test_split(
list(range(len(dataset))), test_size=val_split, random_state=666, shuffle=True
)
return Subset(dataset, train_idx), Subset(da... | @@ -8,7 +8,7 @@
from model_training.custom_datasets.instruction import INSTRUCTION_DATASETS, InstructionDataset
from model_training.custom_datasets.oasst_dataset import load_oasst_export
from model_training.custom_datasets.pretrain_datasets import FanFics, RedPajama
-from model_training.custom_datasets.prompt_dialog... | model/model_training/custom_datasets/__init__.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Dolphin gpt 3.5 data mix | 3,606 | LAION-AI/Open-Assistant | 29,137 | Python | shahules786 | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/__init__.py
PR Title: Dolphin gpt 3.5 data mix
BEFORE CODE:
def train_val_dataset(dataset, val_split=0.2) -> tuple[Dataset, Dataset | None]:
if val_split == 0:
return dataset, None
train_idx, val_idx = tr... | 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 JokeExplaination(Dataset):
name = "joke"
url = "https://gist.github.com/theblackcat102/42b697e24a13fdb499e20edfbf618361/raw/1834dca207898c15f93b809d1195f6f6e47c9e1e/joke_explained.jsonl"
def __init__(self, cache_dir) -> None:
super().__init__()
os.makedirs(cache_dir, exist_ok=True)
... | No issues found. | class JokeExplaination(Dataset):
name = "joke"
url = "https://gist.github.com/theblackcat102/42b697e24a13fdb499e20edfbf618361/raw/1834dca207898c15f93b809d1195f6f6e47c9e1e/joke_explained.jsonl"
def __init__(self, cache_dir) -> None:
super().__init__()
os.makedirs(cache_dir, exist_ok=True)
... | @@ -514,9 +514,8 @@ def process_vicuna_conversations(
def __init__(self, cache_dir: str | Path, mode: str = "sft", input_max_length: int = 32 * 1024) -> None:
super().__init__()
- self.pairs = []
- if mode not in ("sft", "rl"):
- raise NotImplementedError(f"Currently only the mo... | model/model_training/custom_datasets/qa_datasets.py | 0 | Python | 1 | none | 16 | 51 | 51 | true | Changes for orcacode experiment | 3,612 | LAION-AI/Open-Assistant | 29,137 | Python | andreaskoepf | Review this pull request change.
Language: Python
File: model/model_training/custom_datasets/qa_datasets.py
PR Title: Changes for orcacode experiment
BEFORE CODE: class JokeExplaination(Dataset):
name = "joke"
url = "https://gist.github.com/theblackcat102/42b697e24a13fdb499e20edfbf618361/raw/1834dca207898c1... | 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. | |
],
)
elif model_name in family_8b:
return CacheGenConfig(
nlayers=32,
kspecs=[
QuantizationSpec(start_layer=0, end_layer=10, bins=32),
QuantizationSpec(start_layer=10, end_layer=20, bins=16),
... | Is there any support for the default setting (for models other than these five models)? For exmple, we can pass the model_config in vllm (which contains info about the number of layers) or the number of layers directly as the function param, and configure the `CacheGenConfig` dynamically based on this input. | ],
)
elif model_name in family_8b:
return CacheGenConfig(
nlayers=32,
kspecs=[
QuantizationSpec(start_layer=0, end_layer=10, bins=32),
QuantizationSpec(start_layer=10, end_layer=32, bins=16),
... | @@ -39,40 +43,43 @@ def from_model_name(model_name: str) -> "CacheGenConfig":
family_9b = ["THUDM/glm-4-9b-chat"]
if model_name in family_7b:
return CacheGenConfig(
- key_first_layers=10,
- key_second_layers=20,
- key_third_layers=32, # total ... | lmcache/storage_backend/serde/cachegen_basics.py | 26 | Python | 0.571 | performance | 309 | 51 | 51 | false | cachegen config refactor | 156 | LMCache/LMCache | 4,864 | Python | YaoJiayi | Oasis-Git | Review this pull request change.
Language: Python
File: lmcache/storage_backend/serde/cachegen_basics.py
PR Title: cachegen config refactor
BEFORE CODE: ],
)
elif model_name in family_8b:
return CacheGenConfig(
nlayers=32,
kspecs=[
... | Is there any support for the default setting (for models other than these five models)? For exmple, we can pass the model_config in vllm (which contains info about the number of layers) or the number of layers directly as the function param, and configure the `CacheGenConfig` dynamically based on this input. | 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. |
"lmcache_instance_id": {
"type": str,
"default": "lmcache_default_instance",
"env_converter": str,
},
"controller_url": {"type": Optional[str], "default": None, "env_converter": str},
"lmcache_worker_port": {
"type": Optional[int],
"default": None,
"env_co... | @chunxiaozheng Can you explain these two parameters a bit? | "lmcache_instance_id": {
"type": str,
"default": "lmcache_default_instance",
"env_converter": str,
},
"controller_url": {"type": Optional[str], "default": None, "env_converter": str},
"lmcache_worker_port": {
"type": Optional[int],
"default": None,
"env_co... | @@ -142,6 +142,16 @@ def _to_int_list(
"default": None,
"env_converter": int,
},
+ "lmcache_worker_heartbeat_delay_time": {
+ "type": int,
+ "default": 10,
+ "env_converter": int,
+ },
+ "lmcache_worker_heartbeat_time": {
+ "type": Optional[int],
+ "def... | lmcache/v1/config.py | 26 | Python | 0.214 | question | 58 | 51 | 51 | false | [feature] support heartbeat in lmcache worker | 652 | LMCache/LMCache | 4,864 | Python | YaoJiayi | chunxiaozheng | Review this pull request change.
Language: Python
File: lmcache/v1/config.py
PR Title: [feature] support heartbeat in lmcache worker
BEFORE CODE: "lmcache_instance_id": {
"type": str,
"default": "lmcache_default_instance",
"env_converter": str,
},
"controller_url": {"type": Optio... | @chunxiaozheng Can you explain these two parameters a bit? | 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 TYPE_CHECKING:
# First Party
from lmcache.v1.cache_controller.worker import LMCacheWorker
logger = init_logger(__name__)
class LocalCPUBackend(StorageBackendInterface):
"""
The local cpu backend size is variable depending on how much free space is
left in the allocator so we cannot use LRUEvi... | Why can't we leverage the `fmt` argument in the `allocate` function? |
if TYPE_CHECKING:
# First Party
from lmcache.v1.cache_controller.worker import LMCacheWorker
logger = init_logger(__name__)
class LocalCPUBackend(StorageBackendInterface):
"""
The local cpu backend size is variable depending on how much free space is
left in the allocator so we cannot use LRUEvi... | @@ -60,6 +60,7 @@ def __init__(
lookup_server: Optional[LookupServerInterface] = None,
lmcache_worker: Optional["LMCacheWorker"] = None,
layerwise: bool = False,
+ mla: bool = False, | lmcache/v1/storage_backend/local_cpu_backend.py | 26 | Python | 0.357 | question | 68 | 51 | 51 | false | [Refactor][Bugfix] Add KV Cache format storage in local disk backend | 735 | LMCache/LMCache | 4,864 | Python | YaoJiayi | Shaoting-Feng | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/local_cpu_backend.py
PR Title: [Refactor][Bugfix] Add KV Cache format storage in local disk backend
BEFORE CODE:
if TYPE_CHECKING:
# First Party
from lmcache.v1.cache_controller.worker import LMCacheWorker
logger = init_log... | Why can't we leverage the `fmt` argument in the `allocate` function? | 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. |
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# Standard
from dataclasses import dataclass
from typing import Any, List... | Why using any here? Shouldn't it be memory format? | # distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# Future
from __future__ import annotations
# Standard
from dataclasses ... | @@ -32,6 +32,7 @@ class DiskCacheMetadata:
size: int # in bytes
shape: Optional[torch.Size] = None
dtype: Optional[torch.dtype] = None
+ fmt: Any = None | lmcache/utils.py | 26 | Python | 0.429 | question | 50 | 51 | 51 | false | [Refactor][Bugfix] Add KV Cache format storage in local disk backend | 735 | LMCache/LMCache | 4,864 | Python | YaoJiayi | Shaoting-Feng | Review this pull request change.
Language: Python
File: lmcache/utils.py
PR Title: [Refactor][Bugfix] Add KV Cache format storage in local disk backend
BEFORE CODE: # distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See th... | Why using any here? Shouldn't it be memory format? | 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. |
f"Connection initialized/re-established at {self.config.remote_url}"
)
except Exception as e:
with self.lock:
self.failure_time = time.time()
logger.warning(f"Failed to initialize/re-establish remote connection: {e}")
self.connectio... | Why do we remove the error handling here? | return
try:
assert self.config.remote_url is not None
self.connection = CreateConnector(
self.config.remote_url,
self.loop,
self.local_cpu_backend,
self.config,
self.metadata,
)
... | @@ -126,19 +126,7 @@ def contains(self, key: CacheEngineKey, pin: bool = False) -> bool:
logger.warning("Connection is None in contains, returning False")
return False
- future = asyncio.run_coroutine_threadsafe(
- self.connection.exists(key), self.loop
- )
- ... | lmcache/v1/storage_backend/remote_backend.py | 26 | Python | 0.286 | question | 41 | 51 | 51 | false | [feature] add exists sync interface | 828 | LMCache/LMCache | 4,864 | Python | YaoJiayi | chunxiaozheng | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/remote_backend.py
PR Title: [feature] add exists sync interface
BEFORE CODE: f"Connection initialized/re-established at {self.config.remote_url}"
)
except Exception as e:
with self.lock... | Why do we remove the error handling 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. |
self.loop,
self.local_cpu_backend,
self.config,
)
logger.info(
f"Connection initialized/re-established at {self.config.remote_url}"
)
except Exception as e:
with self.lock:
self.failur... | @chunxiaozheng Would you like to add an boolean type extra_config as a switch? | self.loop,
self.local_cpu_backend,
self.config,
self.metadata,
)
logger.info(
f"Connection initialized/re-established at {self.config.remote_url}"
)
except Exception as e:
with self.lo... | @@ -130,12 +130,8 @@ def contains(self, key: CacheEngineKey, pin: bool = False) -> bool:
key.fmt, key.model_name, key.world_size, 0, key.chunk_hash
)
- future = asyncio.run_coroutine_threadsafe( | lmcache/v1/storage_backend/remote_backend.py | 26 | Python | 0.357 | question | 78 | 51 | 51 | false | [feature] add exists sync interface | 828 | LMCache/LMCache | 4,864 | Python | maobaolong | chunxiaozheng | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/remote_backend.py
PR Title: [feature] add exists sync interface
BEFORE CODE: self.loop,
self.local_cpu_backend,
self.config,
)
logger.info(
f"Con... | @chunxiaozheng Would you like to add an boolean type extra_config as a switch? | 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. |
###################
# Scheduler side APIs
####################
@_lmcache_nvtx_annotate
def get_num_new_matched_tokens(
self,
request: "Request",
num_computed_tokens: int,
) -> int:
"""
Check for external KV cache hit.
Args:
request (R... | Directly change `token_ids` instead of deepcopying `request` | ###################
# Scheduler side APIs
####################
@_lmcache_nvtx_annotate
def get_num_new_matched_tokens(
self,
request: "Request",
num_computed_tokens: int,
) -> int:
"""
Check for external KV cache hit.
Args:
request (R... | @@ -786,6 +805,10 @@ def get_num_new_matched_tokens(
the number of tokens that can be loaded from the
external KV cache beyond what is already computed.
"""
+
+ # Preprocess the request for multimodal hashes.
+ request = mask_mm_hashes_in_request(request)
+
if s... | lmcache/integration/vllm/vllm_v1_adapter.py | 26 | Python | 0.429 | suggestion | 60 | 51 | 51 | false | [Core] Support multimodal models that use `mm_hashes` in vLLM | 882 | LMCache/LMCache | 4,864 | Python | YaoJiayi | Shaoting-Feng | Review this pull request change.
Language: Python
File: lmcache/integration/vllm/vllm_v1_adapter.py
PR Title: [Core] Support multimodal models that use `mm_hashes` in vLLM
BEFORE CODE: ###################
# Scheduler side APIs
####################
@_lmcache_nvtx_annotate
def get_num_new_matched... | Directly change `token_ids` instead of deepcopying `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. |
meta.add_request(req_meta)
for request in scheduler_output.scheduled_cached_reqs:
request_tracker = self._request_trackers[request.req_id]
request_tracker.update(request)
req_meta = ReqMeta.from_request_tracker(
request_tracker,
... | remove redundant calls to `mask_mm_hashes_in_request` | for request in scheduler_output.scheduled_cached_reqs:
request_tracker = self._request_trackers[request.req_id]
request_tracker.update(request)
req_meta = ReqMeta.from_request_tracker(
request_tracker,
self._block_size,
self._l... | @@ -925,6 +955,9 @@ def request_finished(
request: "Request",
block_ids: list[int],
) -> tuple[bool, Optional[dict[str, Any]]]:
+ # Preprocess the request for multimodal hashes.
+ request = mask_mm_hashes_in_request(request) | lmcache/integration/vllm/vllm_v1_adapter.py | 26 | Python | 0.429 | suggestion | 53 | 39 | 34 | false | [Core] Support multimodal models that use `mm_hashes` in vLLM | 882 | LMCache/LMCache | 4,864 | Python | YaoJiayi | Shaoting-Feng | Review this pull request change.
Language: Python
File: lmcache/integration/vllm/vllm_v1_adapter.py
PR Title: [Core] Support multimodal models that use `mm_hashes` in vLLM
BEFORE CODE: meta.add_request(req_meta)
for request in scheduler_output.scheduled_cached_reqs:
request_trac... | remove redundant calls to `mask_mm_hashes_in_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. |
if self._mla_worker_id_as0_mode:
key = CacheEngineKey(
key.fmt, key.model_name, key.world_size, 0, key.chunk_hash
)
t1 = time.perf_counter()
future = asyncio.run_coroutine_threadsafe(self.connection.get(key), self.loop)
try:
memory_obj... | If the connection is `None`, are we planning to re-establish the connection? | return None
t2 = time.perf_counter()
self.stats_monitor.update_interval_remote_time_to_get_sync((t2 - t1) * 1000)
if memory_obj is None:
return None
decompressed_memory_obj = self.deserializer.deserialize(memory_obj)
t3 = time.perf_counter()
logge... | @@ -268,6 +268,93 @@ def get_non_blocking(
) -> Optional[Future]:
raise NotImplementedError
+ def batched_get_blocking(
+ self,
+ keys: List[CacheEngineKey],
+ ) -> List[Optional[MemoryObj]]:
+ if self.connection is None:
+ logger.warning("Connection is None in batc... | lmcache/v1/storage_backend/remote_backend.py | 26 | Python | 0.357 | question | 76 | 51 | 51 | false | [feature] support batched_get_blocking in remote backend | 1,027 | LMCache/LMCache | 4,864 | Python | YaoJiayi | chunxiaozheng | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/remote_backend.py
PR Title: [feature] support batched_get_blocking in remote backend
BEFORE CODE: if self._mla_worker_id_as0_mode:
key = CacheEngineKey(
key.fmt, key.model_name, key.world_size, 0, ... | If the connection is `None`, are we planning to re-establish the connection? | 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 NixlChannel:
"""Provides the primitives to send the data and process the received data.
It will have some internal threads to handle the data receiving.
"""
def __init__(
self,
nixl_config: NixlConfigXpYd,
config: LMCacheEngineConfig,
backend: StorageBackendInter... | initialize tp_rank for both sender and receiver | self._receiver_nixl_wrapper.close()
class NixlChannel:
"""Provides the primitives to send the data and process the received data.
It will have some internal threads to handle the data receiving.
"""
def __init__(
self,
nixl_config: NixlConfigXpYd,
config: LMCacheEngine... | @@ -684,7 +687,13 @@ def __init__(
if nixl_config.role == NixlRole.SENDER:
self._sender = NixlSender(nixl_config, config, backend)
else:
- self._receiver = NixlReceiver(nixl_config, config, backend)
+ tp_rank = get_tensor_model_parallel_rank() | lmcache/v1/storage_backend/connector/nixl_connector_v3.py | 26 | Python | 0.071 | suggestion | 47 | 51 | 51 | false | [PD] Support tp for pd | 1,039 | LMCache/LMCache | 4,864 | Python | YaoJiayi | Shaoting-Feng | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/connector/nixl_connector_v3.py
PR Title: [PD] Support tp for pd
BEFORE CODE:
class NixlChannel:
"""Provides the primitives to send the data and process the received data.
It will have some internal threads to handle the dat... | initialize tp_rank for both sender and receiver | 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. |
multiprocessing.set_start_method("spawn", force=True)
self.lookup_server: Optional[LookupServerInterface] = None
if self.enable_p2p:
self.lookup_server = RedisLookupServer(config)
# avoid circular import
# First Party
from lmcache.v1.cache_controller import ... | nitpick: it's safer to check both conditions here: config.enable_nixl and config.nixl_role == "receiver" | multiprocessing.set_start_method("spawn", force=True)
self.lookup_server: Optional[LookupServerInterface] = None
if self.enable_p2p:
self.lookup_server = RedisLookupServer(config)
# avoid circular import
# First Party
from lmcache.v1.cache_controller import ... | @@ -119,7 +119,9 @@ def __init__(
)
# HACK: remove this in the future
- self.remove_after_retrieve = config.enable_nixl
+ # NOTE (Jiayi): This is currently used to support
+ # dropping the kv cache in nixl backend at decoder.
+ self.remove_after_retrieve = config.nixl_rol... | lmcache/v1/cache_engine.py | 26 | Python | 0.357 | nitpick | 104 | 51 | 51 | false | [Core] Make PD and offloading compatible | 1,112 | LMCache/LMCache | 4,864 | Python | vladnosiv | YaoJiayi | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [Core] Make PD and offloading compatible
BEFORE CODE: multiprocessing.set_start_method("spawn", force=True)
self.lookup_server: Optional[LookupServerInterface] = None
if self.enable_p2p:
... | nitpick: it's safer to check both conditions here: config.enable_nixl and config.nixl_role == "receiver" | 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. |
)
else:
# First Party
from lmcache.v1.storage_backend.nixl_backend import NixlBackend
storage_backends["NixlBackend"] = NixlBackend.CreateNixlBackend(
config, metadata
)
assert config.nixl_buffer_device is not None
# T... | what are the restrictions here so as not to include all other features? I tested a similar code and did not see any problems when turning on offloading to disk too | )
else:
# First Party
from lmcache.v1.storage_backend.nixl_backend import NixlBackend
storage_backends["NixlBackend"] = NixlBackend.CreateNixlBackend(
config, metadata
)
assert config.nixl_buffer_device is not None
# T... | @@ -63,29 +63,27 @@ def CreateStorageBackends(
storage_backends["NixlBackend"] = NixlBackend.CreateNixlBackend(
config, metadata, memory_allocator
)
- assert config.nixl_buffer_device is not None
- return storage_backends
-
- # First Party
- ... | lmcache/v1/storage_backend/__init__.py | 26 | Python | 0.357 | question | 163 | 51 | 51 | false | [Core] Make PD and offloading compatible | 1,112 | LMCache/LMCache | 4,864 | Python | vladnosiv | YaoJiayi | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/__init__.py
PR Title: [Core] Make PD and offloading compatible
BEFORE CODE: )
else:
# First Party
from lmcache.v1.storage_backend.nixl_backend import NixlBackend
storage_backen... | what are the restrictions here so as not to include all other features? I tested a similar code and did not see any problems when turning on offloading to disk too | 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 dst_device == "cuda":
dst_device = f"cuda:{torch.cuda.current_device()}"
storage_backends: OrderedDict[str, StorageBackendInterface] = OrderedDict()
if config.enable_nixl:
if config.enable_xpyd:
# First Party
from lmcache.v1.storage_backend.nixl_backend_v3 import... | this is a bit unclear: previously, I could have omitted to specify the inclusion of local_cpu and the rest of the engines would have received it anyway as a buffer, now it requires explicit indication. | if dst_device == "cuda":
dst_device = f"cuda:{torch.cuda.current_device()}"
storage_backends: OrderedDict[str, StorageBackendInterface] = OrderedDict()
if config.enable_nixl:
if config.enable_xpyd:
# First Party
from lmcache.v1.storage_backend.nixl_backend_v3 import... | @@ -63,29 +63,27 @@ def CreateStorageBackends(
storage_backends["NixlBackend"] = NixlBackend.CreateNixlBackend(
config, metadata, memory_allocator
)
- assert config.nixl_buffer_device is not None
- return storage_backends
-
- # First Party
- ... | lmcache/v1/storage_backend/__init__.py | 26 | Python | 0.5 | suggestion | 201 | 51 | 51 | false | [Core] Make PD and offloading compatible | 1,112 | LMCache/LMCache | 4,864 | Python | vladnosiv | YaoJiayi | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/__init__.py
PR Title: [Core] Make PD and offloading compatible
BEFORE CODE: if dst_device == "cuda":
dst_device = f"cuda:{torch.cuda.current_device()}"
storage_backends: OrderedDict[str, StorageBackendInterface] = Or... | this is a bit unclear: previously, I could have omitted to specify the inclusion of local_cpu and the rest of the engines would have received it anyway as a buffer, now it requires explicit indication. | 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. |
storage_backends: OrderedDict[str, StorageBackendInterface] = OrderedDict()
if config.enable_nixl:
if config.enable_xpyd:
# First Party
from lmcache.v1.storage_backend.nixl_backend_v3 import NixlBackend
storage_backends["NixlBackend"] = NixlBackend.CreateNixlBackend... | by the way, it seems there is a place for funny behavior: now it is possible to create a decode node with a local CPU cache, since there is no more validation. however, due to deletion when retrieving, such a cache will work 1 time. maybe I don't see any other limiting factors right away | storage_backends: OrderedDict[str, StorageBackendInterface] = OrderedDict()
if config.enable_nixl:
if config.enable_xpyd:
# First Party
from lmcache.v1.storage_backend.nixl_backend_v3 import NixlBackend
storage_backends["NixlBackend"] = NixlBackend.CreateNixlBackend... | @@ -63,29 +63,29 @@ def CreateStorageBackends(
storage_backends["NixlBackend"] = NixlBackend.CreateNixlBackend(
config, metadata, memory_allocator
)
- assert config.nixl_buffer_device is not None
- return storage_backends
-
- # First Party
- ... | lmcache/v1/storage_backend/__init__.py | 26 | Python | 0.429 | performance | 288 | 51 | 51 | false | [Core] Make PD and offloading compatible | 1,112 | LMCache/LMCache | 4,864 | Python | vladnosiv | YaoJiayi | Review this pull request change.
Language: Python
File: lmcache/v1/storage_backend/__init__.py
PR Title: [Core] Make PD and offloading compatible
BEFORE CODE: storage_backends: OrderedDict[str, StorageBackendInterface] = OrderedDict()
if config.enable_nixl:
if config.enable_xpyd:
# Firs... | by the way, it seems there is a place for funny behavior: now it is possible to create a decode node with a local CPU cache, since there is no more validation. however, due to deletion when retrieving, such a cache will work 1 time. maybe I don't see any other limiting factors right away | 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. |
location = self.storage_manager.contains(key)
if location is None:
# TODO(Jiayi): Need to refactor P2P as a storage backend to
# clean up the following code.
if self.enable_p2p:
fu... | The git diff is not so smart, the true diff for this code block is that I just only add a `if not self.save_only_first_rank or self.metadata.is_first_rank():` outside, so i have to add a tab for each line. | reordered_chunks: List[Tuple[CacheEngineKey, MemoryObj, int, int]] = []
if not self._is_passive():
reordered_chunks = self._process_tokens_internal(
tokens,
mask,
ret_mask,
**kwargs,
)
if self.save_only_first... | @@ -429,69 +441,143 @@ def retrieve(
reordered_memory_objs = []
reordered_starts = []
reordered_ends = []
- for start, end, key in self.token_database.process_tokens(
- tokens=tokens, mask=mask
- ):
- assert isinstance(key, CacheEngineKey)
-
- if... | lmcache/v1/cache_engine.py | 26 | Python | 0.571 | suggestion | 205 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | maobaolong | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: location = self.storage_manager.contains(key)
if location is None:
# TODO(Jiayi): Need to... | The git diff is not so smart, the true diff for this code block is that I just only add a `if not self.save_only_first_rank or self.metadata.is_first_rank():` outside, so i have to add a tab for each line. | 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. |
# NOTE: Unix systems use fork by default
multiprocessing.set_start_method("spawn", force=True)
self.lookup_server: Optional[LookupServerInterface] = None
if self.enable_p2p:
self.lookup_server = RedisLookupServer(config)
# avoid circular import
# First Part... | `config.nixl_role == "receiver"` is for making cpu offload and disaggregated prefill compatible. |
self.enable_p2p = config.enable_p2p
self.enable_controller = config.enable_controller
# NOTE: Unix systems use fork by default
multiprocessing.set_start_method("spawn", force=True)
self.lookup_server: Optional[LookupServerInterface] = None
if self.enable_p2p:
... | @@ -119,11 +130,7 @@ def __init__(
)
# HACK: remove this in the future
- # NOTE (Jiayi): This is currently used to support
- # dropping the kv cache in nixl backend at decoder.
- self.remove_after_retrieve = (
- config.enable_nixl and config.nixl_role == "receiver"
- ... | lmcache/v1/cache_engine.py | 26 | Python | 0.357 | suggestion | 96 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | Shaoting-Feng | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE:
# NOTE: Unix systems use fork by default
multiprocessing.set_start_method("spawn", force=True)
self.lookup_server: Optional[Loo... | `config.nixl_role == "receiver"` is for making cpu offload and disaggregated prefill compatible. | 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. |
dtype=torch.uint8,
device=corrected_device,
)
nixl_cpu_mem_allocator = NixlCPUMemoryAllocator()
nixl_cpu_mem_allocator.init_nixl_memory_allocator(
buffer,
torch.Size(metadata.kv_shape),
... | I suggest we set this as default behavior for MLA models |
if self.lmcache_worker is not None:
self.lmcache_worker.close()
self.storage_manager.close()
self.memory_allocator.close()
logger.info("LMCacheEngine closed.")
def _process_tokens_internal(
self,
tokens,
mask,
ret_mask,
**kwarg... | @@ -807,6 +911,19 @@ def _Create_memory_allocator(
return CuFileMemoryAllocator(config.cufile_buffer_size * 1024**2)
max_local_cpu_size = config.max_local_cpu_size
+ if config.extra_config:
+ save_only_first_rank = config.extra_config.get(
+ "save_only_first_rank... | lmcache/v1/cache_engine.py | 26 | Python | 0.214 | suggestion | 56 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | Shaoting-Feng | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: dtype=torch.uint8,
device=corrected_device,
)
nixl_cpu_mem_allocator = NixlCPUMem... | I suggest we set this as default behavior for MLA models | 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. |
# Copyright 2024-2025 LMCache Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in... | Do other rank workers still need to allocate a cpu buffer? |
if config.weka_path is not None or config.gds_path is not None:
assert config.cufile_buffer_size is not None
return CuFileMemoryAllocator(config.cufile_buffer_size * 1024**2)
max_local_cpu_size = config.max_local_cpu_size
save_only_first_rank_default = True if metadata.... | @@ -807,6 +911,19 @@ def _Create_memory_allocator(
return CuFileMemoryAllocator(config.cufile_buffer_size * 1024**2)
max_local_cpu_size = config.max_local_cpu_size
+ if config.extra_config:
+ save_only_first_rank = config.extra_config.get(
+ "save_only_first_rank... | lmcache/v1/cache_engine.py | 26 | Python | 0.214 | question | 58 | 1,007 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | Shaoting-Feng | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: # Copyright 2024-2025 LMCache Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in complian... | Do other rank workers still need to allocate a cpu buffer? | 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 location not in key_mapping:
key_mapping[location] = [key]
start_mapping[location] = [start]
end_mapping[location] = [end]
continue
assert location is not None
key_map... | I think `torch.distributed.broadcast` can be called the same on both the sender and the receiver, so do we still need to check if `self.metadata.is_first_rank()`? | if self.remove_after_retrieve and not self._is_passive():
self.storage_manager.remove(key)
memory_obj.ref_count_down()
retrieved_tokens = torch.sum(ret_mask)
self.stats_monitor.on_retrieve_finished(monitor_req_id, retrieved_tokens)
logger.info(
... | @@ -429,69 +439,143 @@ def retrieve(
reordered_memory_objs = []
reordered_starts = []
reordered_ends = []
- for start, end, key in self.token_database.process_tokens(
- tokens=tokens, mask=mask
- ):
- assert isinstance(key, CacheEngineKey)
-
- if... | lmcache/v1/cache_engine.py | 26 | Python | 0.643 | question | 162 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | Shaoting-Feng | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: if location not in key_mapping:
key_mapping[location] = [key]
start_mapping[location]... | I think `torch.distributed.broadcast` can be called the same on both the sender and the receiver, so do we still need to check if `self.metadata.is_first_rank()`? | 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. |
or len(reordered_keys) != len(reordered_ends)
or len(reordered_keys) != len(reordered_memory_objs)
):
logger.warning(
f"reordered_keys,reordered_starts,reordered_ends,"
f"reordered_memory_objs hav... | I wonder if it’s possible to broadcast after calling LMCache’s kernel to transfer `memory_obj` from CPU to GPU, instead of using `tensor.to`, since it’s faster. |
:param **kwargs: The additional arguments for the storage backend which
will be passed into the gpu_connector.
return: A generator that yields Optional[torch.Tensor]. The tensor will
be the boolean mask indicating which tokens are retrieved and will
only be returned... | @@ -429,69 +439,143 @@ def retrieve(
reordered_memory_objs = []
reordered_starts = []
reordered_ends = []
- for start, end, key in self.token_database.process_tokens(
- tokens=tokens, mask=mask
- ):
- assert isinstance(key, CacheEngineKey)
-
- if... | lmcache/v1/cache_engine.py | 26 | Python | 0.714 | performance | 160 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | Shaoting-Feng | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: or len(reordered_keys) != len(reordered_ends)
or len(reordered_keys) != len(reordered_memory_objs)
... | I wonder if it’s possible to broadcast after calling LMCache’s kernel to transfer `memory_obj` from CPU to GPU, instead of using `tensor.to`, since it’s faster. | 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. |
tokens: Optional[torch.Tensor] = None,
hashes: Optional[List[int]] = None,
offsets: Optional[List[int]] = None,
mask: Optional[torch.Tensor] = None,
**kwargs,
) -> None:
"""Store the tokens/hashes and mask into the cache engine.
:param Optional[torch.Tensor] ... | Can we add a helper function in CacheEngine with a name like `is_passive`:
```
def is_passive(self):
"""A 'passive' CacheEngine means that the node itself will not store/retrieve
the data directly, but from the "active" worker (i.e., rank 0 in MLA)
"""
return self.save_only_fi... | self.post_inited = False
gc.collect()
if not config.py_enable_gc:
gc.disable()
def post_init(self, **kwargs) -> None:
if not self.post_inited:
logger.info("Post-initializing LMCacheEngine")
self.gpu_connector.initialize_kvcaches_ptr(**kwargs)
... | @@ -178,6 +192,9 @@ def store(
:raises: ValueError if the number of Falses in the mask is not a
multiple of the chunk size.
"""
+ if self.save_only_first_rank and not self.metadata.is_first_rank(): | lmcache/v1/cache_engine.py | 26 | Python | 0.786 | performance | 471 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | ApostaC | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: tokens: Optional[torch.Tensor] = None,
hashes: Optional[List[int]] = None,
offsets: Optional[List[int]] = None,
mask: Opt... | Can we add a helper function in CacheEngine with a name like `is_passive`:
```
def is_passive(self):
"""A 'passive' CacheEngine means that the node itself will not store/retrieve
the data directly, but from the "active" worker (i.e., rank 0 in MLA)
"""
return self.save_only_fi... | 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. |
and the Falses will ALWAYS be at the PREFIX of the tensor.
:param **kwargs: The additional arguments for the storage backend which
will be passed into the gpu_connector.
Should include KV cache specific information (e.g., paged KV buffer
and the page tables).
... | Similar to the above suggestion, we can do:
```suggestion
if not self.is_passive():
``` | next(mem_obj_generator)
self.storage_manager.batched_put(keys[layer_id], memory_objs[layer_id])
else:
# If no cache are found, we still need to yield to avoid
# `StopIteration`
for layer_id in range(self.num_layers):
yield
... | @@ -409,77 +426,29 @@ def retrieve(
ret_mask = torch.zeros_like(tokens, dtype=torch.bool, device="cpu")
- key_mapping: Dict[str, List[CacheEngineKey]] = {}
- start_mapping: Dict[str, List[int]] = {}
- end_mapping: Dict[str, List[int]] = {}
-
reordered_keys = []
reorde... | lmcache/v1/cache_engine.py | 26 | Python | 0.714 | suggestion | 100 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | ApostaC | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: and the Falses will ALWAYS be at the PREFIX of the tensor.
:param **kwargs: The additional arguments for the storage backend which
... | Similar to the above suggestion, we can do:
```suggestion
if not self.is_passive():
``` | 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. |
reordered_ends,
ret_mask,
)
if self.save_only_first_rank:
self._broadcast_or_receive_memory_objs(
reordered_keys,
reordered_starts,
reordered_ends,
reordered_memory_objs,
ret_... | ```suggestion
if self.remove_after_retrieve and not self.is_passive()
``` | Should include KV cache specific information (e.g., paged KV buffer
and the page tables).
:return: the boolean mask indicating which tokens are retrieved. The
length of the mask should be the same as the tokens. On CPU.
:raises: ValueError if the number of Falses in... | @@ -491,7 +460,9 @@ def retrieve(
# TODO(Jiayi): Remove the following for loop with batched operations
for key, memory_obj in zip(reordered_keys, reordered_memory_objs, strict=False):
- if self.remove_after_retrieve:
+ if self.remove_after_retrieve and (
+ not se... | lmcache/v1/cache_engine.py | 26 | Python | 0.643 | suggestion | 87 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | ApostaC | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: reordered_ends,
ret_mask,
)
if self.save_only_first_rank:
self._broadcast_or_receive_mem... | ```suggestion
if self.remove_after_retrieve and not self.is_passive()
``` | 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. |
# synchronize the last layer
next(mem_obj_consumer)
retrieved_tokens = torch.sum(ret_mask)
self.stats_monitor.on_retrieve_finished(monitor_req_id, retrieved_tokens)
logger.debug(
f"Retrieved {retrieved_tokens} "
f"out of {num_required_tokens} "
... | ```suggestion
if self.is_passive():
return
``` | mem_obj_consumer = self.gpu_connector.batched_to_gpu(starts, ends, **kwargs)
next(mem_obj_consumer)
to_count_down = []
for layer_id in range(self.num_layers):
tasks = next(get_generator)
assert None not in tasks
yield Non... | @@ -623,6 +594,9 @@ def prefetch(
"""Launch the prefetching process in the storage manager to load the
KV to the local CPU memory
"""
+ if self.save_only_first_rank:
+ if not self.metadata.is_first_rank():
+ return | lmcache/v1/cache_engine.py | 26 | Python | 0.571 | suggestion | 70 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | ApostaC | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE:
# synchronize the last layer
next(mem_obj_consumer)
retrieved_tokens = torch.sum(ret_mask)
self.stats_monitor.on_retrie... | ```suggestion
if self.is_passive():
return
``` | 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 self.enable_p2p:
self.distributed_server.close()
if self.lmcache_worker is not None:
self.lmcache_worker.close()
self.storage_manager.close()
self.memory_allocator.close()
logger.info("LMCacheEngine closed.")
def _process_tokens_internal(
... | Can we add a docstring for this function to briefly describe what this function does, and what side-effect this function will have on the input parameters (`reordered_keys`, `reordered_memory_objs`, ...) | @_lmcache_nvtx_annotate
def clear(
self,
tokens: Optional[Union[torch.Tensor, List[int]]] = None,
locations: Optional[List[str]] = None,
tags: OrderedDict = None, # TODO: need to clean by tags
) -> int:
if self.save_only_first_rank:
if self.metadata.is_fi... | @@ -857,6 +846,163 @@ def close(self) -> None:
logger.info("LMCacheEngine closed.")
+ def _process_tokens_internal(
+ self,
+ tokens,
+ mask,
+ reordered_keys,
+ reordered_memory_objs,
+ reordered_starts,
+ reordered_ends,
+ ret_mask,
+ ): | lmcache/v1/cache_engine.py | 26 | Python | 0.714 | suggestion | 203 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | ApostaC | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: if self.enable_p2p:
self.distributed_server.close()
if self.lmcache_worker is not None:
self.lmcache_worker.clos... | Can we add a docstring for this function to briefly describe what this function does, and what side-effect this function will have on the input parameters (`reordered_keys`, `reordered_memory_objs`, ...) | 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. |
# receive metadata
metadata_dict = self.broadcast_object_fn(
None, self.metadata.first_rank
)
if metadata_dict is None:
logger.warning(
f"rank={self.metadata.worker... | Also, it would be great if we could have a docstring for this function. | logger.warn(
"The cache block is in the storage, but it can't be retrieved"
)
if (
last_failed_block_start is None
or last_failed_block_start < start
):
... | @@ -857,6 +846,163 @@ def close(self) -> None:
logger.info("LMCacheEngine closed.")
+ def _process_tokens_internal(
+ self,
+ tokens,
+ mask,
+ reordered_keys,
+ reordered_memory_objs,
+ reordered_starts,
+ reordered_ends,
+ ret_mask,
+ ):
+ ... | lmcache/v1/cache_engine.py | 26 | Python | 0.286 | suggestion | 71 | 51 | 51 | false | [VLLM|CacheEngine]support only save on rank 0 for DS MLA | 1,128 | LMCache/LMCache | 4,864 | Python | ApostaC | maobaolong | Review this pull request change.
Language: Python
File: lmcache/v1/cache_engine.py
PR Title: [VLLM|CacheEngine]support only save on rank 0 for DS MLA
BEFORE CODE: # receive metadata
metadata_dict = self.broadcast_object_fn(
None, self.metadata.first_ra... | Also, it would be great if we could have a docstring for this function. | 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.