Exact

Exact Union RegExp

Daniel Han 90ac3b7734 Studio: install causal-conv1d and mamba-ssm only when a model needs them (#13861) ... Studio setup stops downloading causal-conv1d and mamba-ssm. They are installed when a model that needs them is first trained or loaded, and Qwen3.5 / Qwen3.6 / Qwen3-Next no longer request causal-conv1d at all. ## Why #13396 added both kernels to every NVIDIA Linux / Windows setup next to flash-attn: about 555 MB of downloads (~620 MB installed) for every user, while only a few model families use them. Studio already installs them on demand when such a model is trained (`core/training/worker.py`) or loaded for chat (`core/inference/worker.py` through `utils/ssm_runtime.ensure_ssm_runtime`), so the setup step only duplicated that. Qwen3.5 / 3.6 / Qwen3-Next do not need causal-conv1d with Unsloth: unsloth-zoo runs their GatedDeltaNet with its vendored FLA kernels and its own Triton causal conv (`unsloth_zoo/temporary_patches/gdn_causal_conv1d.py`, model types `qwen3_5`, `qwen3_5_moe`, `qwen3_next`). Studio still downloaded the CUDA package for them because their transformers modeling files mention `causal_conv1d`. ## What changes - `studio/install_python_stack.py`: the setup-time SSM install is gone from all three places (initial setup, Linux torch repair, Windows torch flavor repair), with its progress step. flash-attn stays where it was. `unsloth install-kernels` still installs every kernel, for offline images. - `studio/backend/utils/ssm_runtime.py`: `ZOO_GDN_MODEL_TYPES` (equal to zoo's list; a test reads zoo's source to keep them equal). Names and configs of those models, including `_text` / `_vision` sub-configs, resolve to no causal-conv1d request, checked before the modeling-source scan. A config that mixes in another model type that needs it still asks for it. Zoo's own kill switch `UNSLOTH_DISABLE_TRITON_CAUSAL_CONV1D=1` brings the old request back. - `core/training/worker.py` and `core/systemone/clef_runtime.py` use the same predicate, so the eager install, the `is_causal_conv1d_available` hook and Clef (a Qwen3.5 backbone) all agree. - Only where zoo's Triton conv can run: a CUDA torch (not ROCm, not CPU), Triton importable, not macOS. This is read from torch's distribution version (`+rocm` / `+cpu` / `+cu`), so no load imports torch to decide. On ROCm, macOS or Windows without triton-windows, Qwen keeps getting causal-conv1d as before. - With setup no longer covering for it, the on-demand install now decides from the model's config, not only its name: `resolved_model_is_ssm` maps `nemotron_h`, `falcon_h1`, `granitemoehybrid`, `bamba`, `jamba`, `zamba`, `zamba2`, `mamba`, `mamba2`, `falcon_mamba` (and their `_text` sub-configs) to mamba-ssm + causal-conv1d, for training and for the inference load of the resolved base. Previously `NVIDIA-Nemotron-Nano-9B-v2`, `granite-4.0-tiny-preview`, renamed local checkpoints, Jamba, Zamba and Bamba never got mamba-ssm on demand. - After an install inside an inference worker that already imported transformers, the cached `is_causal_conv1d_available` / `is_mamba_ssm_available` / `is_mamba_2_ssm_available` and cached hub-kernel misses are cleared, so switching from Qwen3.5 to Nemotron-H in one session gets the fast path. - The mamba-ssm error raised when it cannot be installed (e.g. offline) says to run `unsloth install-kernels` once while online. - `install_kernel(install_without_gpu=)` is gone; setup was its only caller. - Clef still applies the FLA TileLang guard when it skips the install. Unchanged: LFM2, Mamba / Jamba / Zamba / Bamba still get causal-conv1d on demand; Nemotron-H, Falcon-H1 and Granite-4.0-H still get causal-conv1d then mamba-ssm on demand; nothing already installed is removed, and a usable causal_conv1d still wins inside zoo. ## Tests - `tests/python/test_ssm_kernels_install_python_stack.py`: setup has no SSM step; flash-attn is still installed at all three sites; the CLI kernels are still there. - `test_windows_torch_flavor_invariant.py`: the torch progress totals drop by one on Windows and Linux. - `studio/backend/tests/test_ssm_runtime.py`: Qwen3.5 / 3.6 / Next names and configs install nothing, the kill switch restores the request, a mixed config still asks, the Mamba2 hybrids still install both kernels in order, and the model-type list equals zoo's. - 21 of the new cases fail against main, and the follow-up cases (config-based SSM detection, platform check, cache refresh, offline message) fail against this PR's first commit; all pass here. The touched test files and the setup / resolver suites pass (test_ssm_runtime 84, test_training_worker_flash_attn 67, test_install_python_stack 295, test_cuda_repair 284, test_windows_torch_flavor_invariant 54, among others). ## Accuracy and speed without causal-conv1d (Qwen3.5) Tiny Qwen3.5 / Qwen3-Next through Unsloth, bf16: zoo's Triton conv is bit-identical to torch's reference conv (logits, loss, LoRA grads, 5 optimizer steps); the CUDA package differs from both at bf16 rounding (5.9e-3 logits, grads cos 0.99993). Prefill plus cached token-by-token decode matches a full forward equally with and without the package (4.0e-3 vs 3.9e-3). The profiler shows which kernels ran in every arm. Qwen3.5-0.8B bf16 on a Colab L4, torch 2.14.1+cu130, 3 alternating repeats per arm: | | Without causal-conv1d (zoo Triton) | With causal-conv1d (CUDA) | |---|---|---| | LoRA training, batch 4 x 1024, 30 timed steps | 911.5 ms/step (spread 11.2) | 917.5 ms/step (spread 15.9) | | Greedy decode, 256 new tokens, batch 1 | 19.3 tok/s (spread 0.6) | 19.0 tok/s (spread 0.2) | Both differences are inside the run-to-run spread. Without the package, single-token decode uses torch's `conv_depthwise2d`; with it, `causal_conv1d_update`. A tiny NemotronH also runs identically with a pure-Python mamba-ssm (the Mamba2 path never calls `selective_scan_cuda`), and a merged 16-bit export of a NemotronH LoRA works with neither kernel installed. ## Not covered here - `test_decision_native.py` / `test_decision_serving.py` were not collected locally (missing `fastmcp`).		2026-10-15 08:20:03 -07:00
..
assets
auth	Studio: accept sk-unsloth API keys in the x-api-key header (#13842)	2026-10-15 06:55:49 -07:00
core	Studio: install causal-conv1d and mamba-ssm only when a model needs them (#13861)	2026-10-15 08:20:03 -07:00
hub	Studio: stop listing a GGUF's borrowed text encoder and VAE as a model in the picker (#13778)	2026-10-15 06:36:20 -07:00
integrations/blender
loggers	Studio: declare the Hub's /api/engines retry in the log budget (#13682)	2026-10-13 19:16:53 -07:00
models	Studio: generate images through an external connection (#13756)	2026-10-15 05:06:37 -07:00
picker
plugins
requirements	Studio: stop LTX-2 loads crashing Windows with an access violation on 32 GB machines (#13432)	2026-10-13 02:41:26 -07:00
routes	Studio: serve /v1/completions from MLX models (#13796)	2026-10-15 07:47:52 -07:00
state
storage	Studio: re-embed documents after switching the embedding model (#13702)	2026-10-14 12:23:21 -07:00
studio_mcp	Studio MCP: never run server-side tools from the chat tool (#13605)	2026-10-13 09:00:50 -07:00
tests	Studio: install causal-conv1d and mamba-ssm only when a model needs them (#13861)	2026-10-15 08:20:03 -07:00
utils	Studio: install causal-conv1d and mamba-ssm only when a model needs them (#13861)	2026-10-15 08:20:03 -07:00
vendor
__init__.py
_platform_compat.py
cloudflare_tunnel.py
colab.py
lan_access.py	Studio: serve HTTPS from your own certificate (--ssl-certfile / --ssl-keyfile) (#13447)	2026-10-12 20:45:01 -07:00
main.py	Studio: accept sk-unsloth API keys in the x-api-key header (#13842)	2026-10-15 06:55:49 -07:00
mcp_server.py	Studio: let coding agents use Unsloth Studio over MCP with API keys (#13118)	2026-10-13 05:06:16 -07:00
run.py	Studio: stop llama-server crashing on macOS on Ctrl+C (#13782)	2026-10-15 05:48:42 -07:00
startup_banner.py	Studio: serve HTTPS from your own certificate (--ssl-certfile / --ssl-keyfile) (#13447)	2026-10-12 20:45:01 -07:00