[GH-ISSUE #1757] minimax m2.1 deploy issue #1108

New issue

Closed

opened 2026-10-29 14:48:39 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-29 14:48:39 +00:00

Owner

Originally created by @epheien on GitHub (Dec 26, 2025).
Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/1757

Reminder

I have read the above rules and searched the existing issues.

System Info

[2025-12-26 20:03:09] Load weight end. type=MiniMaxM2ForCausalLM, dtype=torch.float16, avail mem=24.30 GB, mem usage=69.77 GB.
[2025-12-26 20:03:09] Using KV cache dtype: torch.float16
[2025-12-26 20:03:09] The available memory for KV cache is 5.45 GB.
[2025-12-26 20:03:09] max_total_tokens=100000 is larger than the profiled value 23063. Use the profiled value instead.
[2025-12-26 20:03:09] KV Cache is allocated. #tokens: 23063, K size: 2.73 GB, V size: 2.73 GB
[2025-12-26 20:03:09] Memory pool end. avail mem=18.64 GB
[2025-12-26 20:03:09] Capture cuda graph begin. This can take up to several minutes. avail mem=18.14 GB
[2025-12-26 20:03:09] Capture cuda graph bs [1, 2, 4, 8, 12, 16]
Capturing batches (bs=16 avail_mem=18.12 GB):   0%|                                                                                                                       | 0/6 [00:00<?, ?it/s][2025-12-26 20:03:10] Using default W8A8 Block FP8 kernel config. Performance might be sub-optimal! Config file not found at /home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/quantization/configs/N=8192,K=3072,device_name=NVIDIA_RTX_PRO_6000_Blackwell_Workstation_Edition,dtype=fp8_w8a8,block_shape=[128, 128].json
[2025-12-26 20:03:12] Using default W8A8 Block FP8 kernel config. Performance might be sub-optimal! Config file not found at /home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/quantization/configs/N=3072,K=6144,device_name=NVIDIA_RTX_PRO_6000_Blackwell_Workstation_Edition,dtype=fp8_w8a8,block_shape=[128, 128].json
Capturing batches (bs=16 avail_mem=18.12 GB):   0%|                                                                                                                       | 0/6 [00:02<?, ?it/s]
[2025-12-26 20:03:12] Scheduler hit an exception: Traceback (most recent call last):
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/scheduler.py", line 2833, in run_scheduler_process
    scheduler = Scheduler(
                ^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/scheduler.py", line 314, in __init__
    self.init_model_worker()
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/scheduler.py", line 452, in init_model_worker
    self.tp_worker = TpModelWorker(
                     ^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/tp_worker.py", line 253, in __init__
    self._model_runner = ModelRunner(
                         ^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/model_runner.py", line 385, in __init__
    self.initialize(min_per_gpu_memory)
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/model_runner.py", line 584, in initialize
    self.init_device_graphs()
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/model_runner.py", line 2670, in init_device_graphs
    self.graph_runner = graph_runners[self.device](self)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 358, in __init__
    self.capture()
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 513, in capture
    _capture_one_stream()
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 500, in _capture_one_stream
    ) = self.capture_one_batch_size(bs, forward, stream_idx)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 719, in capture_one_batch_size
    run_once()
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 706, in run_once
    logits_output_or_pp_proxy_tensors = forward(
                                        ^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 120, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 795, in forward
    hidden_states = self.model(input_ids, positions, forward_batch, input_embeds)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 710, in forward
    hidden_states, residual = layer(
                              ^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 559, in forward
    hidden_states = self.block_sparse_moe(hidden_states, forward_batch)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 200, in forward
    return self.forward_normal(hidden_states)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 210, in forward_normal
    final_hidden_states = self.experts(hidden_states, topk_output)
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/fused_moe_triton/layer.py", line 908, in forward
    return self.forward_impl(hidden_states, topk_output)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/fused_moe_triton/layer.py", line 927, in forward_impl
    combine_input = self.run_moe_core(
                    ^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/fused_moe_triton/layer.py", line 948, in run_moe_core
    return self.quant_method.apply(
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/kt_ep_wrapper.py", line 965, in apply
    self.submit(layer, dispatch_output)
  File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/kt_ep_wrapper.py", line 895, in submit
    self.wrapper.submit_forward(
  File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/kt_kernel/experts_base.py", line 277, in submit_forward
    self.cpu_infer.submit_with_cuda_stream(
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'kt_kernel_ext.CPUInfer' object has no attribute 'submit_with_cuda_stream'

[2025-12-26 20:03:12] Received sigquit from a child process. It usually means the child failed.

OS: ubuntu 24.04
GPU: NVIDIA RTX Pro 6000

Reproduction

$ kt run m2.1

→ 检测硬件配置...
ℹ GPU: NVIDIA RTX PRO 6000 Blackwell Workstation Edition (95.6GB 显存)
ℹ CPU: AMD EPYC 9135 16-Core Processor (32 核心, 1 NUMA 节点)
ℹ 内存: 251GB

→ 检查模型状态...

ℹ 找到多个匹配的模型，请选择:

选择模型
  [1] MiniMax-M2.1 (MiniMaxAI/MiniMax-M2.1)
  [2] DeepSeek-R1-0528 (deepseek-ai/DeepSeek-R1-0528)
  [3] MiniMax-M2 (MiniMaxAI/MiniMax-M2)

请选择:: 1
ℹ 模型路径: /home/eph/depot/models/MiniMax-M2.1

ℹ Using original precision model (no quantization)

ℹ Auto-computed kt-num-gpu-experts: 79 (TP=1, VRAM=95.6GB per GPU)

→ Configuration
  Model: MiniMax-M2.1
  Path: /home/eph/depot/models/MiniMax-M2.1

  GPU Experts: 79 per layer
  CPU Threads (kt-cpuinfer): 25
  NUMA Nodes (kt-threadpool-count): 1
  Tensor Parallel: 1
  Method: FP8
  Attention: flashinfer

  Server: http://0.0.0.0:30000


Launching server with command:

  /home/eph/depot/wsp/kt/kt/bin/python3 -m sglang.launch_server --host 0.0.0.0 --port 30000 --model /home/eph/depot/models/MiniMax-M2.1 --kt-weight-path /home/eph/depot/models/MiniMax-M2.1
--kt-cpuinfer 25 --kt-threadpool-count 1 --kt-num-gpu-experts 79 --kt-method FP8 --kt-gpu-prefill-token-threshold 4096 --attention-backend flashinfer --trust-remote-code --mem-fraction-static
0.8 --chunked-prefill-size 32768 --max-running-requests 16 --max-total-tokens 100000 --watchdog-timeout 3000 --enable-mixed-chunk --tensor-parallel-size 1 --enable-p2p-check
--served-model-name MiniMax-M2.1 --disable-shared-experts-fusion --fp8-gemm-backend triton --tool-call-parser minimax-m2 --reasoning-parser minimax-append-think

Others

No response

Originally created by @epheien on GitHub (Dec 26, 2025). Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/1757 ### Reminder - [x] I have read the above rules and searched the existing issues. ### System Info ``` [2025-12-26 20:03:09] Load weight end. type=MiniMaxM2ForCausalLM, dtype=torch.float16, avail mem=24.30 GB, mem usage=69.77 GB. [2025-12-26 20:03:09] Using KV cache dtype: torch.float16 [2025-12-26 20:03:09] The available memory for KV cache is 5.45 GB. [2025-12-26 20:03:09] max_total_tokens=100000 is larger than the profiled value 23063. Use the profiled value instead. [2025-12-26 20:03:09] KV Cache is allocated. #tokens: 23063, K size: 2.73 GB, V size: 2.73 GB [2025-12-26 20:03:09] Memory pool end. avail mem=18.64 GB [2025-12-26 20:03:09] Capture cuda graph begin. This can take up to several minutes. avail mem=18.14 GB [2025-12-26 20:03:09] Capture cuda graph bs [1, 2, 4, 8, 12, 16] Capturing batches (bs=16 avail_mem=18.12 GB): 0%| | 0/6 [00:00<?, ?it/s][2025-12-26 20:03:10] Using default W8A8 Block FP8 kernel config. Performance might be sub-optimal! Config file not found at /home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/quantization/configs/N=8192,K=3072,device_name=NVIDIA_RTX_PRO_6000_Blackwell_Workstation_Edition,dtype=fp8_w8a8,block_shape=[128, 128].json [2025-12-26 20:03:12] Using default W8A8 Block FP8 kernel config. Performance might be sub-optimal! Config file not found at /home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/quantization/configs/N=3072,K=6144,device_name=NVIDIA_RTX_PRO_6000_Blackwell_Workstation_Edition,dtype=fp8_w8a8,block_shape=[128, 128].json Capturing batches (bs=16 avail_mem=18.12 GB): 0%| | 0/6 [00:02<?, ?it/s] [2025-12-26 20:03:12] Scheduler hit an exception: Traceback (most recent call last): File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/scheduler.py", line 2833, in run_scheduler_process scheduler = Scheduler( ^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/scheduler.py", line 314, in __init__ self.init_model_worker() File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/scheduler.py", line 452, in init_model_worker self.tp_worker = TpModelWorker( ^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/managers/tp_worker.py", line 253, in __init__ self._model_runner = ModelRunner( ^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/model_runner.py", line 385, in __init__ self.initialize(min_per_gpu_memory) File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/model_runner.py", line 584, in initialize self.init_device_graphs() File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/model_runner.py", line 2670, in init_device_graphs self.graph_runner = graph_runners[self.device](self) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 358, in __init__ self.capture() File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 513, in capture _capture_one_stream() File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 500, in _capture_one_stream ) = self.capture_one_batch_size(bs, forward, stream_idx) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 719, in capture_one_batch_size run_once() File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/model_executor/cuda_graph_runner.py", line 706, in run_once logits_output_or_pp_proxy_tensors = forward( ^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 120, in decorate_context return func(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 795, in forward hidden_states = self.model(input_ids, positions, forward_batch, input_embeds) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl return self._call_impl(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl return forward_call(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 710, in forward hidden_states, residual = layer( ^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl return self._call_impl(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl return forward_call(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 559, in forward hidden_states = self.block_sparse_moe(hidden_states, forward_batch) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl return self._call_impl(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl return forward_call(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 200, in forward return self.forward_normal(hidden_states) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/models/minimax_m2.py", line 210, in forward_normal final_hidden_states = self.experts(hidden_states, topk_output) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1775, in _wrapped_call_impl return self._call_impl(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1786, in _call_impl return forward_call(*args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/fused_moe_triton/layer.py", line 908, in forward return self.forward_impl(hidden_states, topk_output) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/fused_moe_triton/layer.py", line 927, in forward_impl combine_input = self.run_moe_core( ^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/fused_moe_triton/layer.py", line 948, in run_moe_core return self.quant_method.apply( ^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/kt_ep_wrapper.py", line 965, in apply self.submit(layer, dispatch_output) File "/home/eph/depot/wsp/kt/sglang/python/sglang/srt/layers/moe/kt_ep_wrapper.py", line 895, in submit self.wrapper.submit_forward( File "/home/eph/depot/wsp/kt/kt/lib/python3.12/site-packages/kt_kernel/experts_base.py", line 277, in submit_forward self.cpu_infer.submit_with_cuda_stream( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'kt_kernel_ext.CPUInfer' object has no attribute 'submit_with_cuda_stream' [2025-12-26 20:03:12] Received sigquit from a child process. It usually means the child failed. ``` OS: ubuntu 24.04 GPU: NVIDIA RTX Pro 6000 ### Reproduction ```text $ kt run m2.1 → 检测硬件配置... ℹ GPU: NVIDIA RTX PRO 6000 Blackwell Workstation Edition (95.6GB 显存) ℹ CPU: AMD EPYC 9135 16-Core Processor (32 核心, 1 NUMA 节点) ℹ 内存: 251GB → 检查模型状态... ℹ 找到多个匹配的模型，请选择: 选择模型 [1] MiniMax-M2.1 (MiniMaxAI/MiniMax-M2.1) [2] DeepSeek-R1-0528 (deepseek-ai/DeepSeek-R1-0528) [3] MiniMax-M2 (MiniMaxAI/MiniMax-M2) 请选择:: 1 ℹ 模型路径: /home/eph/depot/models/MiniMax-M2.1 ℹ Using original precision model (no quantization) ℹ Auto-computed kt-num-gpu-experts: 79 (TP=1, VRAM=95.6GB per GPU) → Configuration Model: MiniMax-M2.1 Path: /home/eph/depot/models/MiniMax-M2.1 GPU Experts: 79 per layer CPU Threads (kt-cpuinfer): 25 NUMA Nodes (kt-threadpool-count): 1 Tensor Parallel: 1 Method: FP8 Attention: flashinfer Server: http://0.0.0.0:30000 Launching server with command: /home/eph/depot/wsp/kt/kt/bin/python3 -m sglang.launch_server --host 0.0.0.0 --port 30000 --model /home/eph/depot/models/MiniMax-M2.1 --kt-weight-path /home/eph/depot/models/MiniMax-M2.1 --kt-cpuinfer 25 --kt-threadpool-count 1 --kt-num-gpu-experts 79 --kt-method FP8 --kt-gpu-prefill-token-threshold 4096 --attention-backend flashinfer --trust-remote-code --mem-fraction-static 0.8 --chunked-prefill-size 32768 --max-running-requests 16 --max-total-tokens 100000 --watchdog-timeout 3000 --enable-mixed-chunk --tensor-parallel-size 1 --enable-p2p-check --served-model-name MiniMax-M2.1 --disable-shared-experts-fusion --fp8-gemm-backend triton --tool-call-parser minimax-m2 --reasoning-parser minimax-append-think ``` ### Others _No response_

bsakaguchi

2026-10-29 14:48:39 +00:00

closed this issue
added the
bug
label

bsakaguchi commented

2026-10-29 14:48:40 +00:00

Author

Owner

@epheien commented on GitHub (Dec 26, 2025):

You need to recompile kt yourself to solve this problem; you cannot use the software packages on pip.

@epheien commented on GitHub (Dec 26, 2025): You need to recompile kt yourself to solve this problem; you cannot use the software packages on pip.

bsakaguchi referenced this issue

2026-10-29 14:53:30 +00:00

[PR #1108] [MERGED] chore: show cmake output in real time during build_ext #1562

bsakaguchi referenced this issue

2026-10-29 14:53:34 +00:00

[PR #1143] [MERGED] feat(build): display limited tail of subprocesses in real time #1568