-
Notifications
You must be signed in to change notification settings - Fork 48
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
[Bug]: Server crashes (TimeoutError: RPC call to execute_model timed out) when max-num-seqs= 4, tp=4
bugSomething isn't workingSomething isn't workingStatus: Open.#856 In torch-spyre/spyre-inference;[CI] Integration tests are all hitting OOM errors
bugSomething isn't workingSomething isn't workingStatus: Open.#854 In torch-spyre/spyre-inference;Remove Redundant Clone calls for FP8 Linear Kernel
enhancementNew feature or requestNew feature or requestStatus: Open.#840 In torch-spyre/spyre-inference;Remove @dynamo.disable from FP8 apply on M=1 decode
enhancementNew feature or requestNew feature or requestStatus: Open.#837 In torch-spyre/spyre-inference;[Feature]: use KV utilization ratio when selecting attention execution path
enhancementNew feature or requestNew feature or requestStatus: Open.#836 In torch-spyre/spyre-inference;Remove Tiling From Spyre Inference
enhancementNew feature or requestNew feature or requestStatus: Open.#835 In torch-spyre/spyre-inference;[Feature]: extend batched-decode fast path to mixed prefill+decode batches
enhancementNew feature or requestNew feature or requestStatus: Open.#831 In torch-spyre/spyre-inference;- Status: Open.#830 In torch-spyre/spyre-inference;
[Feature]: Align multi-modal model implementation with upstream vLLM
enhancementNew feature or requestNew feature or requestStatus: Open.#828 In torch-spyre/spyre-inference;[Feature]: run RMS norm in fp32 instead of fp16
enhancementNew feature or requestNew feature or requestStatus: Open.[Bug]: test_real_metadata_dispatch_does_not_grow_the_cache failing due to "No module named 'tests.attention'"
bugSomething isn't workingSomething isn't workingStatus: Open.#804 In torch-spyre/spyre-inference;[Bug]: int64→float32 typecast unsupported on Spyre backend - breaks RoPE for Granite/Llama/Gemma via vLLM
bugSomething isn't workingSomething isn't workingStatus: Open.#794 In torch-spyre/spyre-inference;