Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings
This repository was archived by the owner on Sep 4, 2025. It is now read-only.

Commit6a91e80

Browse files
committed
review feedback
1 parent82aa358 commit6a91e80

File tree

8 files changed

+53
-59
lines changed

8 files changed

+53
-59
lines changed

‎blog/ai-code-reviews-reclaims-2023-10-05/blog.md‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -161,9 +161,9 @@ elsewhere.
161161

162162
This isn't just about shipping faster—although who doesn't want to do that? It's
163163
also about optimizing your team's time so they can focus on what they do best:
164-
createamazing software. If an AI code review tool can handle the initial pass,
165-
highlighting potential issues with the code, the developer can resolve initial
166-
feedback faster and your human experts can spend their time solving more
164+
createremarkable software. If an AI code review tool can handle the initial
165+
pass,highlighting potential issues with the code, the developer can resolve
166+
initialfeedback faster and your human experts can spend their time solving more
167167
complex, higher-level problems.
168168

169169
In the next section, we'll understand how AI-powered code reviewers offer a more

‎blog/boosting-engineering-efficiency-2023-11-13/blog.md‎

Lines changed: 0 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -28,11 +28,6 @@ reliable friend who never sleeps (because they're code) and knows all the coding
2828
rules. This article explores how these intelligent bots fill the gaps in our
2929
fast-paced, sometimes slow, new world.
3030

31-
```mdx-code-block
32-
import Tabs from '@theme/Tabs';
33-
import TabItem from '@theme/TabItem';
34-
```
35-
3631
<!--truncate-->
3732

3833
##Direct Correlation: Remote Engineering Challenges & AI Solutions

‎blog/coderabbit-deep-dive-2023-08-26/blog.md‎

Lines changed: 7 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -92,7 +92,7 @@ reviewer to ensure code quality. It goes beyond existing linters and static code
9292
analysis tools in uncovering issues and suggesting improvements by providing a
9393
human-like understanding of the objective of the code.
9494

95-
With CodeRabbit, developers get contextaware feedback within minutes, which
95+
With CodeRabbit, developers get context-aware feedback within minutes, which
9696
enables them to make improvements based on best practices and get their code
9797
ready to be merged faster. CodeRabbit also helps reviewers by providing them
9898
with confidence and speed to approve the code faster. Reviewers can use
@@ -135,7 +135,7 @@ context of the pull request and changes across multiple files. Below is a
135135
glimpse into the challenges we faced and the solutions we came up with:
136136

137137
**Context window size:** The LLM models have limited context windows, for
138-
instance,`gpt-3.5-turbo` has a context window of 4K or 16K tokens and`gpt-4`
138+
instance,`gpt-3.5-turbo` has a context window of 4K or 16K tokens and`gpt-4`
139139
has a context window of 8K tokens. This is often insufficient to pack larger
140140
change sets. To circumvent this, we provide various summaries while reviewing
141141
changes to each file and by smartly prioritizing context that is packed in each
@@ -155,15 +155,14 @@ the noise.
155155

156156
**Costs:** While advanced models like`gpt-4` are great in performing complex
157157
tasks, they are several orders of magnitude more expensive than models like
158-
`gpt-3.5-turbo`. We had to design a multi-model approach that uses simpler
159-
models for summarizations, while complex models are used for tasks such as
160-
reviewing code. In addition, simpler models act as a triage filter that
161-
identifies the changes that need to be thoroughly reviewed by more complex
162-
models.
158+
`gpt-3.5-turbo`. We had to design a multimodel approach that uses simpler models
159+
for summarizations, while complex models are used for tasks such as reviewing
160+
code. In addition, simpler models act as a triage filter that identifies the
161+
changes that need to be thoroughly reviewed by more complex models.
163162

164163
**Inaccuracies:** LLMs are not perfect and often return inaccurate results, and
165164
they sometimes even ignore instructions and completely fabricate a response.
166-
Rather than keep fighting the LLMs we wrote layers of sanity checks to fix or
165+
Rather than keep fighting the LLMs, we wrote layers of sanity checks to fix or
167166
hide the inaccuracies from the user.
168167

169168
**Data privacy:** The biggest concern from our users is whether their code is

‎blog/how-we-built-cost-effective-generative-ai-application-2023-12-23/blog.md‎

Lines changed: 14 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -19,7 +19,7 @@ CodeRabbit reviews several thousand pull requests (PRs) daily. We have
1919
innovative client-side request prioritization technique to navigate OpenAI rate
2020
limits. In this blog post, we will explore how we manage to deliver continuous,
2121
in-depth code analysis cost-effectively, while also providing a robust, free
22-
plan to opensource projects.
22+
plan to open-source projects.
2323

2424
<!--truncate-->
2525

@@ -37,10 +37,10 @@ functionalities. CodeRabbit offers the following tiers of service:
3737
of code changes in a PR.
3838

3939
Our vision is to offer an affordable, AI-driven code review service to
40-
developers and organizations of all sizes while supporting the opensource
41-
community. We are particularly mindful of opensource projects, understanding
40+
developers and organizations of all sizes while supporting the open-source
41+
community. We are particularly mindful of open-source projects, understanding
4242
the challenges in reviewing community contributions. Our goal is to reduce the
43-
burden of code reviews for opensource maintainers by improving submission
43+
burden of code reviews for open-source maintainers by improving submission
4444
quality before the review process begins.
4545

4646
CodeRabbit's review process is automatically triggered when a PR is opened in
@@ -64,7 +64,7 @@ gpt-4 models.
6464
6565
Our primary cost driver is using OpenAI's API to generate code review comments.
6666
We will share our cost optimization strategies in the following sections.
67-
Without these optimizations, our free offering to opensource projects would not
67+
Without these optimizations, our free offering to open-source projects would not
6868
be feasible.
6969

7070
Let's take a look at the strategies that helped us optimize the cost and improve
@@ -94,14 +94,14 @@ changes, we save almost 50% on costs.
9494

9595
##2. Rate-limiting: Enforcing Fair Usage
9696

97-
Upon launching our free service for opensource projects, we noticed individual
97+
Upon launching our free service for open-source projects, we noticed individual
9898
developers using it as a coding co-pilot by making hundreds of incremental
9999
commits for continuous feedback. CodeRabbit, designed for thorough code reviews
100100
unlike tools such as GitHub Copilot, incurs high costs when used in this manner.
101101
Therefore, we implemented hourly rate-limits on the number of files and commits
102102
reviewed per user, to control excessive usage without compromising user
103103
experience. These limits vary across different product tiers. For example, we
104-
set more aggressive limits for opensource users compared to trial and paid
104+
set more aggressive limits for open-source users compared to trial and paid
105105
users.
106106

107107
To implement these rate-limits, we evaluated various options for Serverless
@@ -114,23 +114,23 @@ In FluxNinja Aperture, policies are decoupled from application logic through
114114
labels, enabling new policy additions without altering application code. We
115115
apply labels in FluxNinja Aperture, wrap the review workload with its SDK, and
116116
write policies that enforce limits on those labels. For example, we enforce a 3
117-
reviews per hour limit (1 review every 20 minutes) for opensource users,
117+
reviews per hour limit (1 review every 20 minutes) for open-source users,
118118
allowing a burst of 2 back-to-back reviews, as shown in the screenshots below.
119119

120120
Integration with FluxNinja Aperture SDK
121121

122-
![Ratelimiting commits per hour for opensource users](generative-ai-request-flow-cost-saving.png"Rate limiting commits per hour for open source users")
123-
Ratelimiting commits per hour for opensourceusers
122+
![Rate-limiting commits per hour for open-source users](generative-ai-request-flow-cost-saving.png)
123+
_Rate-limiting commits per hour for open-sourceusers_
124124

125-
![Wait time feedback to the user in a comment](rate-limit-message-screenshot.png"Wait time feedback to the user in a comment")
126-
Wait time feedback to the user in acomment
125+
![Wait time feedback to the user in a comment](rate-limit-message-screenshot.png)
126+
_Wait time feedback to the user in acomment_
127127

128128
Given the high cost and capacity constraints of state-of-the-art models such as
129129
gpt-4, rate-limiting is an essential requirement for any AI application. By
130130
implementing fair-usage rate limits, we are saving almost 20% on our costs.
131131

132-
![Rate limit metrics for open source users](rate-limits-metrics-screenshot.png"image_tRate limit metrics for open source usersooltip")
133-
Rate limit metrics for opensourceusers
132+
![Rate limit metrics for open source users](rate-limits-metrics-screenshot.png)
133+
_Rate limit metrics for open-sourceusers_
134134

135135
##3. Caching: Avoid Re-generating Similar Review Comments
136136

‎blog/openai-rate-limits-2023-10-23/blog.md‎

Lines changed: 12 additions & 12 deletions
Original file line numberDiff line numberDiff line change
@@ -115,7 +115,7 @@ the nature of requests.
115115

116116
We were introduced to the[FluxNinja Aperture](https://www.fluxninja.com/) load
117117
management platform by one of our advisors.
118-
[Aperture](https://github.com/fluxninja/aperture) is an opensource load
118+
[Aperture](https://github.com/fluxninja/aperture) is an open-source load
119119
management platform that offers advanced rate-limiting, request prioritization,
120120
and quota management features. Essentially, Aperture serves as a global token
121121
bucket, facilitating client-side rate limits and business-attribute-based
@@ -150,13 +150,13 @@ context to Aperture, we also attach the following labels to each request:
150150
tier. For example, the reasoning could that the user is a`paid_user`,
151151
`trial_user` or a`open_source_user`. Requests to OpenAI are prioritized based
152152
on these labels.
153-
-`priority`: Requests are ranked according a priority number provided in this
154-
label. For instance, requests from`paid_user` are given precedence over those
155-
from`trial_user` and`open_source_user`. The base priority is incremented for
156-
each file reviewed, enabling pull requests that are further along in the
157-
review process to complete more quickly than newly submitted ones.
158-
Additionally, chat messages are assigned a much higher priority compared to
159-
review tasks.
153+
-`priority`: Requests are ranked accordingtoa priority number provided in
154+
thislabel. For instance, requests from`paid_user` are given precedence over
155+
thosefrom`trial_user` and`open_source_user`. The base priority is
156+
incremented foreach file reviewed, enabling pull requests that are further
157+
along in thereview process to complete more quickly than newly submitted
158+
ones.Additionally, chat messages are assigned a much higher priority compared
159+
toreview tasks.
160160

161161
![Request Flow](request-flow.svg)
162162

@@ -227,22 +227,22 @@ context to Aperture, we also attach the following labels to each request:
227227

228228
Aperture offers a foundational "blueprint" for
229229
[managing quotas](https://docs.fluxninja.com/reference/blueprints/quota-scheduling/base),
230-
comprisingoftwo main components:
230+
comprising two main components:
231231

232232
- Rate limiter: OpenAI employs a token bucket algorithm to impose rate limits,
233233
and that is directly compatible with Aperture's rate limiter. For example, in
234234
the tokens per minute policy for`gpt-4`, we have allocated a burst capacity
235235
of`40000 tokens`, and a refill rate of`40000 tokens per minute`. The bucket
236236
begins to refill the moment the tokens are withdrawn, aligning with OpenAI's
237-
ratelimiting mechanism. This ensures our outbound request and token rate
237+
rate-limiting mechanism. This ensures our outbound request and token rate
238238
remains synchronized with OpenAI's enforced limits.
239239
- Scheduler: Aperture has a
240240
[weighted fair queuing](https://docs.fluxninja.com/concepts/scheduler/)
241241
scheduler that prioritizes the requests based on multiple factors such as the
242242
number of tokens, priority levels and workload labels.
243243

244-
By fine-tuning these two components in Aperture, weare able togo as fast as we
245-
can,with optimal user experience, while ensuring that we don't exceed the rate
244+
By fine-tuning these two components in Aperture, wecango as fast as we can,
245+
with optimal user experience, while ensuring that we don't exceed the rate
246246
limits.
247247

248248
<details>

‎docs/about/support.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -29,8 +29,8 @@ which they are purchased and cannot be used under another organization.
2929
###Can CodeRabbit review my existing PRs after integrating it?
3030

3131
CodeRabbit would by default only review new PRs or existing PRs which have a new
32-
commit after the app is installed. You can, however, use`@coderabbitai review`
33-
command on the PR to trigger a review of existing PRs.
32+
commit after the app is installed. You can, however, usethe
33+
`@coderabbitai review`command on the PR to trigger a review of existing PRs.
3434

3535
###Who can add the repositories on CodeRabbit?
3636

‎docs/guides/customize-coderabbit.md‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -8,8 +8,7 @@ description:
88
---
99

1010
CodeRabbit offers various customization options to tailor the reviews to your
11-
specific requirements. Customizations can be made using one of the below options
12-
:
11+
requirements. Customizations can be made using one of the below options :
1312

1413
1.**Customization via CodeRabbit YAML File**_(Recommended)_
1514
2.**Customization via CodeRabbit UI**
@@ -19,7 +18,8 @@ Customization defined in the YAML file overrides the settings in the UI.
1918
###1. Customize CodeRabbit using a YAML File[](https://coderabbit.ai/docs/get-started/customize-coderabbit)
2019

2120
You can add a`.coderabbit.yaml` configuration file to your repositories. Below
22-
is a sample YAML file that can be used as a starting point and change as needed:
21+
is a sample YAML file that can be used as a starting point and changed as
22+
needed:
2323

2424
####version 2:`Current`
2525

@@ -217,14 +217,14 @@ Reviews can also be customized through the UI under repository settings:
217217
**Path-based instructions :** Provide additional review guidelines based on the
218218
file paths using glob patterns.
219219

220-
For further guidance, please refer this
220+
For further guidance, please refertothis
221221
section [Prompt Customization](/guides/prompt-customization)
222222

223223
**Language Selection:** Select the natural language in which you want the review
224224
feedback. CodeRabbit supports most of the widely used languages. The default
225225
language is English.
226226

227-
**Disable High Level Summary:** This allows disabling highlevel summary added
227+
**Disable High Level Summary:** This allows disabling high-level summary added
228228
with the pull request description.
229229

230230
---
@@ -236,7 +236,7 @@ with the pull request description.
236236
**Disable Reviews:** Disables automatic code reviews for the repository. Reviews
237237
can be initiated on-demand using CodeRabbit commands.
238238

239-
**EnableDraft Pull requestReviews:** Select whether CodeRabbit should review
239+
**Enabledraft pull requestreviews:** Select whether CodeRabbit should review
240240
the draft pull requests.
241241

242242
**Configure reviews using pull request labels:** Select specific pull request

‎docs/introduction/introduction.md‎

Lines changed: 9 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -3,19 +3,19 @@ id: introduction
33
title:What is CodeRabbit?
44
sidebar_label:Introduction
55
description:
6-
CodeRabbit is an innovative, AI-first code reviewer. It provides contextaware
6+
CodeRabbit is an innovative, AI-first code reviewer. It provides context-aware
77
review feedback on a pull request within minutes, thereby significantly
88
reducing the time and effort associated with manual code reviews.
99
slug:"/"
1010
---
1111

12-
**CodeRabbit** is an innovative, AI-first code reviewer. It provides context
13-
aware review feedback on a pull request within minutes, thereby significantly
14-
reducing the time and effort associated with manual code reviews. Secondly, it
15-
brings a different perspective to the reviews and finds issues often missed by
16-
the human eye. Developers can chat with the bot within the code, which allows
17-
them to provide additional context, ask questions, or have the bot generate
18-
code. It learns from your suggestions and gets better overtime.
12+
**CodeRabbit** is an innovative, AI-first code reviewer. It provides
13+
context-aware review feedback on a pull request within minutes, thereby
14+
significantlyreducing the time and effort associated with manual code reviews.
15+
Secondly, itbrings a different perspective to the reviews and finds issues
16+
often missed bythe human eye. Developers can chat with the bot within the code,
17+
which allowsthem to provide additional context, ask questions, or have the bot
18+
generatecode. It learns from your suggestions and gets better overtime.
1919

2020
**CodeRabbit** seamlessly integrates with GitHub and GitLab repositories,
2121
performing continuous, incremental reviews for each commit within a pull
@@ -42,7 +42,7 @@ process with other third parties.
4242
Does CodeRabbit store my Data? We store embeddings of your conversations, which
4343
are used to improve future reviews for your organization.
4444

45-
Compliance and Data Privacy: Your data is kept confidential, isolated by
45+
Compliance and Data Privacy: Your data is kept confidential, isolated by the
4646
organization, and used only to improve your organization reviews. Data storage
4747
complies with SOC2 Type II, GDPR, and HIPAA standards.
4848

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp