-
I run `python examples/benchmark.py` and encounter the following error - the output from flex attention differs from SDPA
```
Traceback (most recent call last):
File "/local/home/chayut/workpla…
-
Enable periodic nightly jobs to enable tracking performance changes over time during development.
Inspiration: https://cockroachdb.github.io/pebble/?max=local
marun updated
6 months ago
-
Hi,
I am thoroughly impressed by the performance of Qwen2-VL on the Agent Benchmarks. However, I was unable to locate the corresponding testing code in the repository. Could you please point me to …
-
Performance counters may be helpful for benchmarking:
- it can provide more information like cache misses, branch misprediction, etc.
- number of cycles may be more stable than wall clock time.
I…
-
Try to use data sets mentioned in the [faiss](https://arxiv.org/pdf/2401.08281) paper for testing:
- BIGANN: A commonly used dataset for similarity search benchmarking, containing 128-dimensional S…
-
Hi,
Very interesting work. I have some remarks:
* In your paper, you oftenly speak about "compared to Tensorflow", what kind of distribution strategy do you talk about ? What kind of strategy: x…
-
Hello everyone,
It is exciting to see that the project is now on GitHub!
I wanted to share some thoughts on possible directions to improve both the codebase and future collaboration.
While ther…
-
## What
Plan a round of moderated research and testing for a first prototype of a community space and contribution form. Work alongside the Interaction Designer and Content Designer whilst they develo…
-
## Goal
| User story |
|:---------------------------------------------------------------------------|
| As an endpoint engineer,
| I want Fleet to support the latest macOS CIS Benchmarks
…
-
Hello, thanks for your great work. I have some little questions.
When testing a Qwen2 based model, like `llava_qwen` or `lmms-lab/LongVA-7B`, on V-NIAH benchmark,
there is a function [apply_seq_…