ollama run qwen3-next:80b-a3b-instruct-q4_K_M Error: 500 Internal Server Error: failed to initialize model: qwen3next: layer 0 missing attn_qkv/attn_gate projections

Fix Plan

Fix Name: Update Model Configuration

Step 1: Update Model Configuration File

Update the ollama.toml file to include the missing attn_qkv and attn_gate projections for the qwen3-next model.

[model.qwen3-next]
layer0.attn_qkv = true
layer0.attn_gate = true

Step 2: Update Model Code (if necessary)

If the model code is not automatically generated, update the qwen3-next model code to include the missing projections.

# qwen3-next.py
import torch
import torch.nn as nn

class Qwen3NextLayer0(nn.Module):
    def __init__(self):
        super(Qwen3NextLayer0, self).__init__()
        self.attn_qkv = nn.Linear(128, 128 * 3)  # Update projection size
        self.attn_gate = nn.Linear(128, 128)  # Update projection size

    def forward(self, x):
        # ...
        attn_qkv = self.attn_qkv(x)
        attn_gate = self.attn_gate(x)
        # ...

Step 3: Re-run Ollama

Run ollama run qwen3-next:80b-a3b-instruct-q4_K_M again to verify the fix.

Verification

• Check the Ollama logs for any errors.
• Verify that the model is running without errors.
• Test the model with sample inputs to ensure it produces the expected output.