Spaces:

---

Major-TOM

/

MajorTOM-Core-Viewer

like 81

Running on L4

Hi @mikonvergence
I have upgraded Python to 3.9 for leafmap compatibility and optimized duckdb configuration to resolve timeout issues on HF Spaces.
If any issues persist, I’ll be happy to look into them further.

Best,
Harsh

That's fantastic Harsh, thank you!

Let me see if the new text prompt functionality works now.

Great the tweak worked! @mikonvergence

If there’s anything else you need, just let me know. I’d also be happy to contribute to any projects.

Also, since Python 3.9 is nearing end-of-support, do you think we should update the Docker environment to Python 3.10?

Amazing, thanks!

In fact, there does still seem to be an issue when text query is tested.

Faiss assertion 'err == CUBLAS_STATUS_SUCCESS' failed in void faiss::gpu::runBatchMatrixMult(faiss::gpu::Tensor<float, 3, true>&, bool, faiss::gpu::Tensor<U, 3, true, int, faiss::gpu::traits::DefaultPtrTraits>&, bool, faiss::gpu::Tensor<BT, 3, true>&, bool, float, float, cublasHandle_t, cudaStream_t) [with AT = float; BT = float; cublasHandle_t = cublasContext*; cudaStream_t = CUstream_st*] at /project/faiss/faiss/gpu/utils/MatrixMult-inl.cuh:355; details: cublasGemmStridedBatchedEx failed (13)

I would be happy to upgrade to 3.10, but I remember 3.11 was problematic given some of the libraries used in the project related to the open_clip library (if I remember correctly).

I'll try to find a solution, it seems to be related to:
https://github.com/facebookresearch/faiss/issues/2064

If you manage to fix in the meantime, let me know!