Issues
- 3
assert self.d_ssm % self.headdim == 0
#360 opened by songxujay - 0
- 7
- 13
RuntimeError: causal_conv1d with channel last layout requires strides (x.stride(0) and x.stride(2)) to be multiples of 8
#351 opened by xiaoxudaxfcv - 8
Mamba-2: IndexError: map::at
#361 opened by Prophet-Kathleen - 0
The simple test of the model works fine, but there is an "Aborted (core dumped)" issue during training.
#375 opened by xueaa - 3
训练时loss.backward出现问题了
#373 opened by WDYTX - 0
How to abandon the use of z?
#374 opened by xuepengcheng1231 - 8
Trition Error
#370 opened by Kevin-naticl - 2
triton error
#365 opened by missingthl - 3
triton error while running Mamba2 with slow path
#369 opened by Seeker98 - 0
Alternative implementation of Multi-Head Mamba
#372 opened by vidavakil - 10
- 4
Bug in selective scan backward
#368 opened by Hprairie - 9
Tips on debugging CUDA kernel
#339 opened by Hprairie - 24
Loss NaN in Mamba2
#352 opened by tyshiwo1 - 6
AssertionError exclusive to Mamba2
#347 opened by Madiwka4 - 3
ImportError: /home/antony/MambaTrack/.venv/lib/python3.8/site-packages/selective_scan_cuda.cpython-38-x86_64-linux-gnu.so: undefined symbol: _ZN2at4_ops10zeros_like4callERKNS_6TensorEN3c108optionalINS5_10ScalarTypeEEENS6_INS5_6LayoutEEENS6_INS5_6DeviceEEENS6_IbEENS6_INS5_12MemoryFormatEEE
#358 opened by AhJayzZ - 2
Why mamba2 is much slower than mamba?
#367 opened by dwgan - 1
Mamba2 using Triton 2.1.0 error: FileNotFoundError: [Errno 2] No such file or directory: 'ldconfig'
#363 opened by Mang30 - 1
Mamba2 Parallelization questioned. Very similar to the 'UltraLight VM-UNet: Parallel Vision Mamba Significantly Reduces Parameters for Skin Lesion Segmentation' article.
#364 opened by Nobuttero - 4
- 2
[Mamba2] Loss is all NaN during training
#353 opened by XiudingCai - 2
Chunk Size and Nan Loss
#354 opened by bio-mlhui - 0
- 19
Error when trying to use Mamba2
#345 opened by yxchng - 2
where is the src.ops.triton.k_activations?
#349 opened by gaotai8 - 1
Small Mistake in Mamba2 - Eq (9)
#350 opened by vasqu - 1
- 1
No module named k_activations
#346 opened by Yiwen233 - 10
Missing Gradient in Fast_Path Mode?
#342 opened by Leopold2333 - 2
- 2
is mamba2 released?
#344 opened by yxchng - 2
BLOCK_N in _layer_norm_fwd()
#340 opened by vidavakil - 4
- 2
access hidden_states of MambaLMHeadModel
#337 opened by s6mahahn - 1
mamba_ssm安装成功,导入失败,无法使用。
#338 opened by Lijuming33 - 6
A question about matrix A
#326 opened by WaitDumplings - 2
Question about loading data?
#336 opened by Hprairie - 4
- 1
mamba-ssm for macos with M1
#324 opened by freedom-cognit - 2
import SSM problem
#331 opened by zhanke199423 - 4
Training with step
#333 opened by norikazu99 - 1
- 8
ImportError
#330 opened by talkinglim - 3
WarpReverseScan Exclusive Scan Bug
#327 opened by Hprairie - 2
- 2
Different scan algorithm used in forward and backward selective scan kernel
#329 opened by minhkhoi1026 - 2
RuntimeError while trying to run MambaLMHeadModel
#325 opened by ph-marra - 3
What should I do?Critical libmamba [json.exception.type_error.302] type must be string, but is null
#322 opened by helloworldABCD1234