Not working on linux with rtx 5050 8gb

[Zapptix]

python corridorkey_cli.py wizard /home/zapp/CorridorKey/ClipsForInference/myshot/
[03:10:47] INFO     Auto-selected device: cuda                                                   device_utils.py:22
           INFO     Using device: cuda                                                       corridorkey_cli.py:218
╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ CORRIDOR KEY — SMART WIZARD                                                                                     │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Windows Path: /home/zapp/CorridorKey/ClipsForInference/myshot/
Running locally: /home/zapp/CorridorKey/ClipsForInference/myshot/

Found 1 potential clip folders.
                Status Report                 
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━┳━━━━━━━┓
┃ Category                   ┃ Count ┃ Clips ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━╇━━━━━━━┩
│ Ready (AlphaHint)          │     1 │ —     │
├────────────────────────────┼───────┼───────┤
│ Masked (VideoMaMaMaskHint) │     0 │ —     │
├────────────────────────────┼───────┼───────┤
│ Raw (Input only)           │     0 │ —     │
└────────────────────────────┴───────┴───────┘
╭──────────────────────────────────────────────────── Actions ────────────────────────────────────────────────────╮
│ i — Run Inference (1 ready clips)                                                                               │
│ r — Re-scan folders                                                                                             │
│ q — Quit                                                                                                        │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Select action [v/g/b/i/r/q] (q): i
╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ Corridor Key Inference                                                                                          │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ Inference Settings                                                                                              │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Input colorspace [linear/srgb] (srgb): 
Despill strength (0–10, 10 = max despill) (5): 
Enable auto-despeckle (removes tracking dots)? [y/n] (y): 
Despeckle size (min pixels for a spot) (400): 
Refiner strength multiplier (experimental) (1.0): 
[03:10:57] INFO     Found 1 clips ready for inference.                                          clip_manager.py:612
           INFO     Not Apple Silicon — using torch backend                                           backend.py:53
           INFO     Torch engine loaded: CorridorKey.pth (device=cuda)                               backend.py:238
           INFO     Loading CorridorKey from                                                 inference_engine.py:68
                    /home/zapp/CorridorKey/CorridorKeyModule/checkpoints/CorridorKey.pth                           
           INFO     Initializing hiera_base_plus_224.mae_in1k_ft_in1k (img_size=2048)      model_transformer.py:159
[03:10:58] INFO     Skipped downloading base weights (relying on custom checkpoint)        model_transformer.py:164
           INFO     Patched input layer: 3 → 4 channels (extra initialized to 0)           model_transformer.py:240
           INFO     Feature channels: [112, 224, 448, 896]                                 model_transformer.py:177
In file included from /home/zapp/.local/share/uv/python/cpython-3.13.12-linux-x86_64-gnu/include/python3.13/Python.h:14,
                 from /tmp/tmp611v2r9t/__triton_launcher.c:7:
/home/zapp/.local/share/uv/python/cpython-3.13.12-linux-x86_64-gnu/include/python3.13/pyconfig.h:1967:9: warning: ‘_POSIX_C_SOURCE’ redefined
 1967 | #define _POSIX_C_SOURCE 200809L
      |         ^~~~~~~~~~~~~~~
In file included from /usr/include/bits/libc-header-start.h:33,
                 from /usr/include/stdlib.h:26,
                 from /home/zapp/CorridorKey/.venv/lib/python3.13/site-packages/triton/backends/nvidia/include/cuda.h:56,
                 from /tmp/tmp611v2r9t/__triton_launcher.c:2:
/usr/include/features.h:319:10: note: this is the location of the previous definition
  319 | # define _POSIX_C_SOURCE        202405L
      |          ^~~~~~~~~~~~~~~
In file included from /home/zapp/.local/share/uv/python/cpython-3.13.12-linux-x86_64-gnu/include/python3.13/Python.h:14,
                 from /tmp/tmp0fipe8x2/__triton_launcher.c:7:
/home/zapp/.local/share/uv/python/cpython-3.13.12-linux-x86_64-gnu/include/python3.13/pyconfig.h:1967:9: warning: ‘_POSIX_C_SOURCE’ redefined
 1967 | #define _POSIX_C_SOURCE 200809L
      |         ^~~~~~~~~~~~~~~
In file included from /usr/include/bits/libc-header-start.h:33,
                 from /usr/include/stdlib.h:26,
                 from /home/zapp/CorridorKey/.venv/lib/python3.13/site-packages/triton/backends/nvidia/include/cuda.h:56,
                 from /tmp/tmp0fipe8x2/__triton_launcher.c:2:
/usr/include/features.h:319:10: note: this is the location of the previous definition
  319 | # define _POSIX_C_SOURCE        202405L
      |          ^~~~~~~~~~~~~~~
[03:11:06] INFO     Running Inference on:                                                       clip_manager.py:631
           INFO       Input frames: 1, Alpha frames: 1 -> Processing 1 frames                   clip_manager.py:646
⠏  ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 0/1 0:00:00
Inference failed: Dynamo failed to run FX node with fake tensors: call_function <built-in method conv2d of type 
object at 0x7fc79672e9e0>(*(FakeTensor(..., device='cuda:0', size=(1, s27, 2048, 2048)), Parameter(FakeTensor(..., 
device='cuda:0', size=(112, 4, 7, 7), requires_grad=True)), Parameter(FakeTensor(..., device='cuda:0', size=(112,),
requires_grad=True)), (4, 4), (3, 3), (1, 1), 1), **{}): got RuntimeError('Given groups=1, weight of size [112, 4, 
7, 7], expected input[1, s27, 2048, 2048] to have 4 channels, but got s27 channels instead')

from user code:
   File "/home/zapp/CorridorKey/CorridorKeyModule/core/model_transformer.py", line 247, in forward
    features = self.encoder(x)  # Returns list of features
  File "/home/zapp/CorridorKey/.venv/lib/python3.13/site-packages/timm/models/_features.py", line 476, in forward
    features = self.model.forward_intermediates(
  File "/home/zapp/CorridorKey/.venv/lib/python3.13/site-packages/timm/models/hiera.py", line 758, in 
forward_intermediates
    x = self.patch_embed(x, mask=patch_mask)
  File "/home/zapp/CorridorKey/.venv/lib/python3.13/site-packages/timm/models/hiera.py", line 455, in forward
    x = self.proj(x)
  File "/home/zapp/CorridorKey/.venv/lib/python3.13/site-packages/torch/nn/modules/conv.py", line 553, in forward
    return self._conv_forward(input, self.weight, self.bias)
  File "/home/zapp/CorridorKey/.venv/lib/python3.13/site-packages/torch/nn/modules/conv.py", line 548, in 
_conv_forward
    return F.conv2d(

Set TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a bug to 
PyTorch). For even more developer context, set TORCH_LOGS="+dynamo"

Inference batch complete. Press Enter to re-scan: 

Im using Linux with hyprland on arch. my system has RTX 5050 with 8gb of vram, 16gb of ddr4 ram

[Zapptix]

Starting Corridor Key locally...
Target: "C:\Users\Prash\Documents\Sample"
Uninstalled 2 packages in 1.95s
Installed 2 packages in 20.76s
[22:03:02] INFO     Auto-selected device: cpu                                                                                    device_utils.py:22
           INFO     Using device: cpu                                                                                        corridorkey_cli.py:218
╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ CORRIDOR KEY — SMART WIZARD                                                                                                                     │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Windows Path: C:\Users\Prash\Documents\Sample
Running locally: C:\Users\Prash\Documents\Sample

Found 1 potential clip folders.
                 Status Report
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━┳━━━━━━━━┓
┃ Category                   ┃ Count ┃ Clips  ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━╇━━━━━━━━┩
│ Ready (AlphaHint)          │     1 │ Sample │
├────────────────────────────┼───────┼────────┤
│ Masked (VideoMaMaMaskHint) │     0 │ —      │
├────────────────────────────┼───────┼────────┤
│ Raw (Input only)           │     0 │ —      │
└────────────────────────────┴───────┴────────┘
╭──────────────────────────────────────────────────────────────────── Actions ────────────────────────────────────────────────────────────────────╮
│ i — Run Inference (1 ready clips)                                                                                                               │
│ r — Re-scan folders                                                                                                                             │
│ q — Quit                                                                                                                                        │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Select action [v/g/b/i/r/q] (q): i
╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ Corridor Key Inference                                                                                                                          │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
╭─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│ Inference Settings                                                                                                                              │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
Input colorspace [linear/srgb] (srgb):
Despill strength (0–10, 10 = max despill) (5):
Enable auto-despeckle (removes tracking dots)? [y/n] (y):
Despeckle size (min pixels for a spot) (400):
Refiner strength multiplier (experimental) (1.0):
[22:03:17] INFO     Found 1 clips ready for inference.                                                                          clip_manager.py:612
           INFO     Not Apple Silicon — using torch backend                                                                           backend.py:53
           INFO     Torch engine loaded: CorridorKey.pth (device=cpu)                                                                backend.py:238
           INFO     Loading CorridorKey from                                                                                 inference_engine.py:68
                    C:\Users\Prash\Downloads\CorridorKey-main\CorridorKey-main\CorridorKeyModule\checkpoints\CorridorKey.pth
           INFO     Initializing hiera_base_plus_224.mae_in1k_ft_in1k (img_size=2048)                                      model_transformer.py:159
[22:03:18] INFO     Skipped downloading base weights (relying on custom checkpoint)                                        model_transformer.py:164
           INFO     Patched input layer: 3 → 4 channels (extra initialized to 0)                                           model_transformer.py:240
           INFO     Feature channels: [112, 224, 448, 896]                                                                 model_transformer.py:177
[22:03:36] INFO     Model compilation failed with error: RuntimeError: Compiler: cl is not found.                            inference_engine.py:62

                    Set TORCHDYNAMO_VERBOSE=1 for the internal stack trace (please do this especially if you're reporting a
                    bug to PyTorch). For even more developer context, set TORCH_LOGS="+dynamo"

           WARNING  Model compilation failed. Falling back to eager mode.                                                    inference_engine.py:63
           INFO     Running Inference on: Sample                                                                                clip_manager.py:631
           INFO       Input frames: 1, Alpha frames: 1 -> Processing 1 frames                                                   clip_manager.py:646
⠹ Sample ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 0/1 0:00:00
Inference failed: Given groups=1, weight of size [112, 4, 7, 7], expected input[1, 7, 2048, 2048] to have 4 channels, but got 7 channels instead
Inference batch complete. Press Enter to re-scan:

I tried this one on Windows 11 as well with the same system and still am getting this error