vllm
oasdiff
vllm | oasdiff | |
---|---|---|
31 | 12 | |
19,344 | 587 | |
12.6% | 5.3% | |
9.9 | 9.2 | |
2 days ago | 5 days ago | |
Python | Go | |
Apache License 2.0 | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
vllm
-
AI leaderboards are no longer useful. It's time to switch to Pareto curves
I guess the root cause of my claim is that OpenAI won't tell us whether or not GPT-3.5 is an MoE model, and I assumed it wasn't. Since GPT-3.5 is clearly nondeterministic at temp=0, I believed the nondeterminism was due to FPU stuff, and this effect was amplified with GPT-4's MoE. But if GPT-3.5 is also MoE then that's just wrong.
What makes this especially tricky is that small models are truly 100% deterministic at temp=0 because the relative likelihoods are too coarse for FPU issues to be a factor. I had thought 3.5 was big enough that some of its token probabilities were too fine-grained for the FPU. But that's probably wrong.
On the other hand, it's not just GPT, there are currently floating-point difficulties in vllm which significantly affect the determinism of any model run on it: https://github.com/vllm-project/vllm/issues/966 Note that a suggested fix is upcasting to float32. So it's possible that GPT-3.5 is using an especially low-precision float and introducing nondeterminism by saving money on compute costs.
Sadly I do not have the money[1] to actually run a test to falsify any of this. It seems like this would be a good little research project.
[1] Or the time, or the motivation :) But this stuff is expensive.
-
Mistral AI Launches New 8x22B Moe Model
The easiest is to use vllm (https://github.com/vllm-project/vllm) to run it on a Couple of A100's, and you can benchmark this using this library (https://github.com/EleutherAI/lm-evaluation-harness)
- FLaNK AI for 11 March 2024
-
Show HN: We got fine-tuning Mistral-7B to not suck
Great question! scheduling workloads onto GPUs in a way where VRAM is being utilised efficiently was quite the challenge.
What we found was the IO latency for loading model weights into VRAM will kill responsiveness if you don't "re-use" sessions (i.e. where the model weights remain loaded and you run multiple inference sessions over the same loaded weights).
Obviously projects like https://github.com/vllm-project/vllm exist but we needed to build out a scheduler that can run a fleet of GPUs for a matrix of text/image vs inference/finetune sessions.
disclaimer: I work on Helix
-
Mistral CEO confirms 'leak' of new open source AI model nearing GPT4 performance
FYI, vLLM also just added experiment multi-lora support: https://github.com/vllm-project/vllm/releases/tag/v0.3.0
Also check out the new prefix caching, I see huge potential for batch processing purposes there!
- VLLM Sacrifices Accuracy for Speed
- Easy, fast, and cheap LLM serving for everyone
- vllm
- Mixtral Expert Parallelism
- Mixtral 8x7B Support
oasdiff
- FLaNK AI for 11 March 2024
-
Serverless APIs: Auto-Generate OpenAPI Docs & CI/CD Protections
We will use an open-source GitHub action, oasdiff-action, based on the tool ‘oasdiff.’
-
How Can You Achieve Continuous Deployment for *APIs*?
Nice, Have you come across this tool oasdiff from the article? It may help with detect API breaking changes in swagger
-
How to prevent breaking API changes with API Gateway
While you might wish that pull request reviewers would spot any breaking changes, relying solely on this method is not certain and might lead to failure eventually. If you have OpenAPI/Swagger documentation for your APIs, these can be version-controlled and included in a CI pipeline. APISIX doesn't natively support direct integration with version control systems like Git for API specification changes. However, you can set up a process outside APISIX. Tools like Oasdiff or Bump can identify changes in API specs, and trigger a CI pipeline (add GitHub Action) that runs tests against the route endpoints in APISIX to ensure no breaking changes are introduced.
- Would you like to be notified when your API provider makes a breaking change?
-
Testing for Breaking Changes in Fastify APIs
Now that we have a way to lookup our API’s behavior with Git, we can start testing for breaking changes between versions of our API. We’ll be using Optic (an open source tool I created) to do just that. If you are looking for other options I recommend https://github.com/OpenAPITools/openapi-diff or https://github.com/Tufin/oasdiff.
-
Ask HN: Non-Breaking API deprecation in OpenAPI spec – what do you think?
2. Delete the API at the sunset date or later
People seem to want such a process in order to prevent breaking-changes.
I wrote a diff tool for OpenAPI spec which supports detection of breaking-changes and I recently extended it to support this process and a bit more.
Now I'm looking for feedback.
Proposed Solution (currently in Beta): https://github.com/Tufin/oasdiff#non-breaking-removal-of-deprecated-resources
Related requests:
- A diff tool and Go module for OpenAPI Specification
- OpenAPI Diff
What are some alternatives?
TensorRT - NVIDIA® TensorRT™ is an SDK for high-performance deep learning inference on NVIDIA GPUs. This repository contains the open source components of TensorRT.
openapi-preprocessor - An authoring tool for OpenAPI specifications
CTranslate2 - Fast inference engine for Transformer models
apiclarity - An API security tool to capture and analyze API traffic, test API endpoints, reconstruct Open API specification, and identify API security risks.
lmdeploy - LMDeploy is a toolkit for compressing, deploying, and serving LLMs.
openapi-diff - Utility for comparing two OpenAPI specifications.
Llama-2-Onnx
openapi-generator-go - An opinionated OpenAPI v3 code generator for Go. Use this to generate API models and router scaffolding.
tritony - Tiny configuration for Triton Inference Server
api-firewall - Fast and light-weight API proxy firewall for request and response validation by OpenAPI specs.
faster-whisper - Faster Whisper transcription with CTranslate2
Optic - OpenAPI linting, diffing and testing. Optic helps prevent breaking changes, publish accurate documentation and improve the design of your APIs.