[GH-ISSUE #1111] [Bug] AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux' #775

Open
opened 2026-10-29 14:44:19 +00:00 by bsakaguchi · 10 comments
Owner

Originally created by @iKpuc on GitHub (Apr 10, 2025).
Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/1111

Checklist

  • 1. I have searched related issues but cannot get the expected help.
  • 2. The bug has not been fixed in the latest version.
  • 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.
  • 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/kvcache-ai/ktransformers/discussions. Otherwise, it will be closed.
  • 5. To help the community, I will use Chinese/English or attach an Chinese/English translation if using another language. Non-Chinese/English content without translation may be closed.

Describe the bug

Hello!

The build was created without problems, however, an error occurs when launching Local_chat:

$ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL

2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend
found flashinfer
using custom modeling_xxx.py.
Traceback (most recent call last):
File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in
fire.Fire(local_chat)
File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire
component_trace = _Fire(component, args, parsed_flag_args, context, name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire
component, remaining_args = _CallAndUpdateTrace(
^^^^^^^^^^^^^^^^^^^^
File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace
component = fn(*varargs, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^
File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat
model = custom_modelsconfig.architectures[0]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init
self.model = DeepseekV3Model(config)
^^^^^^^^^^^^^^^^^^^^^^^
File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init
[
File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in
DeepseekV3DecoderLayer(config, layer_idx)
File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init
DeepseekV3MoE(config)
File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init
self.gate = MoEGate(config)
^^^^^^^^^^^^^^^
File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init
self.seq_aux = config.seq_aux
^^^^^^^^^^^^^^
File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute
return super().getattribute(key)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux'

Reproduction

python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL

Environment

Ubuntu 22.04
2*Xeon 8592+
512gb Ddr5

Originally created by @iKpuc on GitHub (Apr 10, 2025). Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/1111 ### Checklist - [x] 1. I have searched related issues but cannot get the expected help. - [x] 2. The bug has not been fixed in the latest version. - [x] 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback. - [x] 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/kvcache-ai/ktransformers/discussions. Otherwise, it will be closed. - [x] 5. To help the community, I will use Chinese/English or attach an Chinese/English translation if using another language. Non-Chinese/English content without translation may be closed. ### Describe the bug Hello! The build was created without problems, however, an error occurs when launching Local_chat: $ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL 2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend found flashinfer using custom modeling_xxx.py. Traceback (most recent call last): File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in fire.Fire(local_chat) File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire component, remaining_args = _CallAndUpdateTrace( ^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat model = custom_modelsconfig.architectures[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init self.model = DeepseekV3Model(config) ^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init [ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in DeepseekV3DecoderLayer(config, layer_idx) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init DeepseekV3MoE(config) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init self.gate = MoEGate(config) ^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init self.seq_aux = config.seq_aux ^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute return super().getattribute(key) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux' ### Reproduction python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL ### Environment Ubuntu 22.04 2*Xeon 8592+ 512gb Ddr5
Author
Owner

@iKpuc commented on GitHub (Apr 10, 2025):

The dependencies are installed as in flash-infer, I tried to run it through the API, the same thing...

<!-- gh-comment-id:2794311990 --> @iKpuc commented on GitHub (Apr 10, 2025): The dependencies are installed as in flash-infer, I tried to run it through the API, the same thing...
Author
Owner

@KMSorSMS commented on GitHub (Apr 11, 2025):

Checklist

  • 1. I have searched related issues but cannot get the expected help.[x] 2. The bug has not been fixed in the latest version.[x] 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.[x] 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/kvcache-ai/ktransformers/discussions. Otherwise, it will be closed.[x] 5. To help the community, I will use Chinese/English or attach an Chinese/English translation if using another language. Non-Chinese/English content without translation may be closed.

Describe the bug

Hello!

The build was created without problems, however, an error occurs when launching Local_chat:

$ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL

2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend found flashinfer using custom modeling_xxx.py. Traceback (most recent call last): File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in fire.Fire(local_chat) File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire component, remaining_args = _CallAndUpdateTrace( ^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat model = custom_modelsconfig.architectures[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init self.model = DeepseekV3Model(config) ^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init [ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in DeepseekV3DecoderLayer(config, layer_idx) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init DeepseekV3MoE(config) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init self.gate = MoEGate(config) ^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init self.seq_aux = config.seq_aux ^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute return super().getattribute(key) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux'

Reproduction

python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL

Environment

Ubuntu 22.04 2*Xeon 8592+ 512gb Ddr5

Could you please show your flash infer version, pip list | grep flash and pip show flashinfer-python, and check if it is v0.2.3? As we use our own flash_infer in ./third_party/custom_flashinfer, you may need to install it instead of the official release. Run pip install third_party/custom_flashinfer/ may help you.

<!-- gh-comment-id:2795685251 --> @KMSorSMS commented on GitHub (Apr 11, 2025): > ### Checklist > * [x] 1. I have searched related issues but cannot get the expected help.[x] 2. The bug has not been fixed in the latest version.[x] 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.[x] 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/kvcache-ai/ktransformers/discussions. Otherwise, it will be closed.[x] 5. To help the community, I will use Chinese/English or attach an Chinese/English translation if using another language. Non-Chinese/English content without translation may be closed. > > ### Describe the bug > Hello! > > The build was created without problems, however, an error occurs when launching Local_chat: > > $ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL > > 2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend found flashinfer using custom modeling_xxx.py. Traceback (most recent call last): File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in fire.Fire(local_chat) File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire component, remaining_args = _CallAndUpdateTrace( ^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat model = custom_modelsconfig.architectures[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init self.model = DeepseekV3Model(config) ^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init [ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in DeepseekV3DecoderLayer(config, layer_idx) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init DeepseekV3MoE(config) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init self.gate = MoEGate(config) ^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init self.seq_aux = config.seq_aux ^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute return super().getattribute(key) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux' > > ### Reproduction > python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL > > ### Environment > Ubuntu 22.04 2*Xeon 8592+ 512gb Ddr5 Could you please show your flash infer version, `pip list | grep flash` and ` pip show flashinfer-python`, and check if it is v0.2.3? As we use our own flash_infer in ./third_party/custom_flashinfer, you may need to install it instead of the official release. Run `pip install third_party/custom_flashinfer/` may help you.
Author
Owner

@iKpuc commented on GitHub (Apr 11, 2025):

Checklist

  • 1. I have searched related issues but cannot get the expected help.[x] 2. The bug has not been fixed in the latest version.[x] 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.[x] 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/kvcache-ai/ktransformers/discussions. Otherwise, it will be closed.[x] 5. To help the community, I will use Chinese/English or attach an Chinese/English translation if using another language. Non-Chinese/English content without translation may be closed.

Describe the bug

Hello!
The build was created without problems, however, an error occurs when launching Local_chat:
$ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL
2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend found flashinfer using custom modeling_xxx.py. Traceback (most recent call last): File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in fire.Fire(local_chat) File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire component, remaining_args = _CallAndUpdateTrace( ^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat model = custom_modelsconfig.architectures[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init self.model = DeepseekV3Model(config) ^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init [ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in DeepseekV3DecoderLayer(config, layer_idx) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init DeepseekV3MoE(config) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init self.gate = MoEGate(config) ^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init self.seq_aux = config.seq_aux ^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute return super().getattribute(key) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux'

Reproduction

python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL

Environment

Ubuntu 22.04 2*Xeon 8592+ 512gb Ddr5

Could you please show your flash infer version, pip list | grep flash and pip show flashinfer-python, and check if it is v0.2.3? As we use our own flash_infer in ./third_party/custom_flashinfer, you may need to install it instead of the official release. Run pip install third_party/custom_flashinfer/ may help you.

Hi! I did what you wrote. It didn't help....

(ktransformers) user@user:/ktransformers$ pip list | grep flash flashinfer-python 0.2.3
(ktransformers) user@user:
/ktransformers$ pip show flashinfer-python Name: flashinfer-python
Version: 0.2.3
Summary: FlashInfer: Kernel Library for LLM Serving
Home-page: https://github.com/flashinfer-ai/flashinfer
Author: FlashInfer team
Author-email:
License: Apache License 2.0
Location: /home/user/anaconda3/envs/ktransformers/lib/python3.11/site-packages
Requires: ninja, numpy, torch
Required-by:

<!-- gh-comment-id:2797843501 --> @iKpuc commented on GitHub (Apr 11, 2025): > > ### Checklist > > > > > > * [x] 1. I have searched related issues but cannot get the expected help.[x] 2. The bug has not been fixed in the latest version.[x] 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.[x] 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/kvcache-ai/ktransformers/discussions. Otherwise, it will be closed.[x] 5. To help the community, I will use Chinese/English or attach an Chinese/English translation if using another language. Non-Chinese/English content without translation may be closed. > > ### Describe the bug > > Hello! > > The build was created without problems, however, an error occurs when launching Local_chat: > > $ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL > > 2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend found flashinfer using custom modeling_xxx.py. Traceback (most recent call last): File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in fire.Fire(local_chat) File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire component, remaining_args = _CallAndUpdateTrace( ^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat model = custom_modelsconfig.architectures[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init self.model = DeepseekV3Model(config) ^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init [ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in DeepseekV3DecoderLayer(config, layer_idx) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init DeepseekV3MoE(config) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init self.gate = MoEGate(config) ^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init self.seq_aux = config.seq_aux ^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute return super().getattribute(key) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux' > > ### Reproduction > > python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL > > ### Environment > > Ubuntu 22.04 2*Xeon 8592+ 512gb Ddr5 > > Could you please show your flash infer version, `pip list | grep flash` and ` pip show flashinfer-python`, and check if it is v0.2.3? As we use our own flash_infer in ./third_party/custom_flashinfer, you may need to install it instead of the official release. Run `pip install third_party/custom_flashinfer/` may help you. Hi! I did what you wrote. It didn't help.... (ktransformers) user@user:~/ktransformers$ pip list | grep flash flashinfer-python 0.2.3 (ktransformers) user@user:~/ktransformers$ pip show flashinfer-python Name: flashinfer-python Version: 0.2.3 Summary: FlashInfer: Kernel Library for LLM Serving Home-page: https://github.com/flashinfer-ai/flashinfer Author: FlashInfer team Author-email: License: Apache License 2.0 Location: /home/user/anaconda3/envs/ktransformers/lib/python3.11/site-packages Requires: ninja, numpy, torch Required-by:
Author
Owner

@KMSorSMS commented on GitHub (Apr 12, 2025):

Seems like this issue is related to our latest fix. May I check your installed version? Try the git log and also show it with pip list | grep ktrans. If older, maybe you can try reinstalling the latest version.

<!-- gh-comment-id:2798767165 --> @KMSorSMS commented on GitHub (Apr 12, 2025): Seems like this issue is related to our latest fix. May I check your installed version? Try the `git log` and also show it with `pip list | grep ktrans`. If older, maybe you can try reinstalling the latest version.
Author
Owner

@iKpuc commented on GitHub (Apr 12, 2025):

Seems like this issue is related to our latest fix. May I check your installed version? Try the git log and also show it with pip list | grep ktrans. If older, maybe you can try reinstalling the latest version.

commit 7a19f3b781 (HEAD)
Merge: 85e2cc7 22df52e
Author: wang jiahao 44601537+qiyuxinlin@users.noreply.github.com
Date: Thu Feb 27 21:01:21 2025 +0800

Merge pull request #721 from kvcache-ai/fix_temperature

fix temperature
<!-- gh-comment-id:2798803234 --> @iKpuc commented on GitHub (Apr 12, 2025): > Seems like this issue is related to our latest fix. May I check your installed version? Try the `git log` and also show it with `pip list | grep ktrans`. If older, maybe you can try reinstalling the latest version. commit 7a19f3b78116391b0c809285997e7b8df1ba5c0d (HEAD) Merge: 85e2cc7 22df52e Author: wang jiahao <44601537+qiyuxinlin@users.noreply.github.com> Date: Thu Feb 27 21:01:21 2025 +0800 Merge pull request #721 from kvcache-ai/fix_temperature fix temperature
Author
Owner

@iKpuc commented on GitHub (Apr 12, 2025):

Seems like this issue is related to our latest fix. May I check your installed version? Try the git log and also show it with pip list | grep ktrans. If older, maybe you can try reinstalling the latest version.

When Building commit 6617549, the following problem occurs:
(venv) (base) user@user:/ktransformers$ export USE_NUMA=1
(venv) (base) user@user:
/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation
Using Python 3.11.11 environment at: venv
× Failed to build ktransformers @ file:///home/user/ktransformers
├─▶ The build backend returned an error
╰─▶ Call to setuptools.build_meta.prepare_metadata_for_build_wheel failed (exit status: 1)

  [stderr]
  Traceback (most recent call last):
    File "<string>", line 14, in <module>
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel
      self.run_setup()
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup
      exec(code, locals())
    File "<string>", line 445, in <module>
  ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set.

  hint: This usually indicates a problem with the package or the build environment.
<!-- gh-comment-id:2798811450 --> @iKpuc commented on GitHub (Apr 12, 2025): > Seems like this issue is related to our latest fix. May I check your installed version? Try the `git log` and also show it with `pip list | grep ktrans`. If older, maybe you can try reinstalling the latest version. When Building commit 6617549, the following problem occurs: (venv) (base) user@user:~/ktransformers$ export USE_NUMA=1 (venv) (base) user@user:~/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation Using Python 3.11.11 environment at: venv × Failed to build `ktransformers @ file:///home/user/ktransformers` ├─▶ The build backend returned an error ╰─▶ Call to `setuptools.build_meta.prepare_metadata_for_build_wheel` failed (exit status: 1) [stderr] Traceback (most recent call last): File "<string>", line 14, in <module> File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel self.run_setup() File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup exec(code, locals()) File "<string>", line 445, in <module> ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set. hint: This usually indicates a problem with the package or the build environment.
Author
Owner

@iKpuc commented on GitHub (Apr 12, 2025):

Seems like this issue is related to our latest fix. May I check your installed version? Try the git log and also show it with pip list | grep ktrans. If older, maybe you can try reinstalling the latest version.

And there is the same error with this commit.:
commit c51818c39a (HEAD, tag: v0.2.3post2)
Merge: bda9cf1 3934b9d
Author: Atream 80757050+Atream@users.noreply.github.com
Date: Sat Mar 15 23:09:30 2025 +0800

Merge pull request #902 from kvcache-ai/rollback-triton-prefill

rollback-triton-prefill

$ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL
2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend found flashinfer using custom modeling_xxx.py. Traceback (most recent call last): File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in fire.Fire(local_chat) File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire component, remaining_args = _CallAndUpdateTrace( ^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat model = custom_modelsconfig.architectures[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init self.model = DeepseekV3Model(config) ^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init [ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in DeepseekV3DecoderLayer(config, layer_idx) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init DeepseekV3MoE(config) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init self.gate = MoEGate(config) ^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init self.seq_aux = config.seq_aux ^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute return super().getattribute(key) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux'

<!-- gh-comment-id:2798817024 --> @iKpuc commented on GitHub (Apr 12, 2025): > Seems like this issue is related to our latest fix. May I check your installed version? Try the `git log` and also show it with `pip list | grep ktrans`. If older, maybe you can try reinstalling the latest version. And there is the same error with this commit.: commit c51818c39a980baaa754b4306564385d2596fd3b (HEAD, tag: v0.2.3post2) Merge: bda9cf1 3934b9d Author: Atream <80757050+Atream@users.noreply.github.com> Date: Sat Mar 15 23:09:30 2025 +0800 Merge pull request #902 from kvcache-ai/rollback-triton-prefill rollback-triton-prefill > $ python ./ktransformers/local_chat.py --model_path deepseek-ai/DeepSeek-R1 --gguf_path ~/DeepSeek-R1-UD-Q2_K_XL/ --cpu_infer 65 --max_new_tokens 8192 --temperature 0.6 --model_name unsloth/DeepSeek-R1-UD-Q2_K_XL 2025-04-09 21:00:11,184 - INFO - flashinfer.jit: Prebuilt kernels not found, using JIT backend found flashinfer using custom modeling_xxx.py. Traceback (most recent call last): File "/home/user/ktransformers/./ktransformers/local_chat.py", line 183, in fire.Fire(local_chat) File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 135, in Fire component_trace = _Fire(component, args, parsed_flag_args, context, name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 468, in _Fire component, remaining_args = _CallAndUpdateTrace( ^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/fire/core.py", line 684, in _CallAndUpdateTrace component = fn(*varargs, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/local_chat.py", line 91, in local_chat model = custom_modelsconfig.architectures[0] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1605, in init self.model = DeepseekV3Model(config) ^^^^^^^^^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1367, in init [ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1368, in DeepseekV3DecoderLayer(config, layer_idx) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 1154, in init DeepseekV3MoE(config) File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 515, in init self.gate = MoEGate(config) ^^^^^^^^^^^^^^^ File "/home/user/ktransformers/./ktransformers/models/modeling_deepseek_v3.py", line 401, in init self.seq_aux = config.seq_aux ^^^^^^^^^^^^^^ File "/home/user/ktransformers/venv/lib/python3.11/site-packages/transformers/configuration_utils.py", line 264, in getattribute return super().getattribute(key) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'DeepseekV3Config' object has no attribute 'seq_aux'
Author
Owner

@KMSorSMS commented on GitHub (Apr 12, 2025):

Seems like this issue is related to our latest fix. May I check your installed version? Try the git log and also show it with pip list | grep ktrans. If older, maybe you can try reinstalling the latest version.

When Building 6617549, the following problem occurs: (venv) (base) user@user:/ktransformers$ export USE_NUMA=1 (venv) (base) user@user:/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation Using Python 3.11.11 environment at: venv × Failed to build ktransformers @ file:///home/user/ktransformers ├─▶ The build backend returned an error ╰─▶ Call to setuptools.build_meta.prepare_metadata_for_build_wheel failed (exit status: 1)

  [stderr]
  Traceback (most recent call last):
    File "<string>", line 14, in <module>
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel
      self.run_setup()
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup
      exec(code, locals())
    File "<string>", line 445, in <module>
  ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set.

  hint: This usually indicates a problem with the package or the build environment.
  1. Your CUDA_HOME may not set. You can echo $CUDA_HOME to see if it is correct and export the correct path.
  2. You may use the latest commit to check which is: 35ba63e
  3. But the quickest way to check is to download an older version of deepseek-ai/DeepSeekV3 config locally. Because the config file on the official net is updated, you can find an older one that fits in with our older version to see if this is the problem. And if it is, you can go ahead to use our latest version as it updates the new Deepseek v3 config support.
<!-- gh-comment-id:2798817403 --> @KMSorSMS commented on GitHub (Apr 12, 2025): > > Seems like this issue is related to our latest fix. May I check your installed version? Try the `git log` and also show it with `pip list | grep ktrans`. If older, maybe you can try reinstalling the latest version. > > When Building [6617549](https://github.com/kvcache-ai/ktransformers/commit/6617549de0d1f40b0e9254105be7dcc11bf97de5), the following problem occurs: (venv) (base) user@user:~/ktransformers$ export USE_NUMA=1 (venv) (base) user@user:~/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation Using Python 3.11.11 environment at: venv × Failed to build `ktransformers @ file:///home/user/ktransformers` ├─▶ The build backend returned an error ╰─▶ Call to `setuptools.build_meta.prepare_metadata_for_build_wheel` failed (exit status: 1) > > ``` > [stderr] > Traceback (most recent call last): > File "<string>", line 14, in <module> > File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel > self.run_setup() > File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup > exec(code, locals()) > File "<string>", line 445, in <module> > ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set. > > hint: This usually indicates a problem with the package or the build environment. > ``` 1. Your CUDA_HOME may not set. You can echo $CUDA_HOME to see if it is correct and export the correct path. 2. You may use the latest commit to check which is: [35ba63e](https://github.com/kvcache-ai/ktransformers/tree/35ba63e259913ad16a1e49ad8a277235a43eb555) 3. But the quickest way to check is to download an older version of deepseek-ai/DeepSeekV3 config locally. Because the config file on the official net is updated, you can find an older one that fits in with our older version to see if this is the problem. And if it is, you can go ahead to use our latest version as it updates the new Deepseek v3 config support.
Author
Owner

@iKpuc commented on GitHub (Apr 12, 2025):

Seems like this issue is related to our latest fix. May I check your installed version? Try the git log and also show it with pip list | grep ktrans. If older, maybe you can try reinstalling the latest version.

When Building 6617549, the following problem occurs: (venv) (base) user@user:/ktransformers$ export USE_NUMA=1 (venv) (base) user@user:/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation Using Python 3.11.11 environment at: venv × Failed to build ktransformers @ file:///home/user/ktransformers ├─▶ The build backend returned an error ╰─▶ Call to setuptools.build_meta.prepare_metadata_for_build_wheel failed (exit status: 1)

  [stderr]
  Traceback (most recent call last):
    File "<string>", line 14, in <module>
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel
      self.run_setup()
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup
      exec(code, locals())
    File "<string>", line 445, in <module>
  ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set.

  hint: This usually indicates a problem with the package or the build environment.
  1. Your CUDA_HOME may not set. You can echo $CUDA_HOME to see if it is correct and export the correct path.
  2. You may use the latest commit to check which is: 35ba63e
  3. But the quickest way to check is to download an older version of deepseek-ai/DeepSeekV3 config locally. Because the config file on the official net is updated, you can find an older one that fits in with our older version to see if this is the problem. And if it is, you can go ahead to use our latest version as it updates the new Deepseek v3 config support.

Thanks for the help! However, after creating the build, the following error occurred:
Image

<!-- gh-comment-id:2798892757 --> @iKpuc commented on GitHub (Apr 12, 2025): > > > Seems like this issue is related to our latest fix. May I check your installed version? Try the `git log` and also show it with `pip list | grep ktrans`. If older, maybe you can try reinstalling the latest version. > > > > > > When Building [6617549](https://github.com/kvcache-ai/ktransformers/commit/6617549de0d1f40b0e9254105be7dcc11bf97de5), the following problem occurs: (venv) (base) user@user:~/ktransformers$ export USE_NUMA=1 (venv) (base) user@user:~/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation Using Python 3.11.11 environment at: venv × Failed to build `ktransformers @ file:///home/user/ktransformers` ├─▶ The build backend returned an error ╰─▶ Call to `setuptools.build_meta.prepare_metadata_for_build_wheel` failed (exit status: 1) > > ``` > > [stderr] > > Traceback (most recent call last): > > File "<string>", line 14, in <module> > > File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel > > self.run_setup() > > File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup > > exec(code, locals()) > > File "<string>", line 445, in <module> > > ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set. > > > > hint: This usually indicates a problem with the package or the build environment. > > ``` > > 1. Your CUDA_HOME may not set. You can echo $CUDA_HOME to see if it is correct and export the correct path. > 2. You may use the latest commit to check which is: [35ba63e](https://github.com/kvcache-ai/ktransformers/tree/35ba63e259913ad16a1e49ad8a277235a43eb555) > 3. But the quickest way to check is to download an older version of deepseek-ai/DeepSeekV3 config locally. Because the config file on the official net is updated, you can find an older one that fits in with our older version to see if this is the problem. And if it is, you can go ahead to use our latest version as it updates the new Deepseek v3 config support. Thanks for the help! However, after creating the build, the following error occurred: ![Image](https://github.com/user-attachments/assets/b250122e-bd4f-4fa5-baf4-570e2154b0fb)
Author
Owner

@KMSorSMS commented on GitHub (Apr 13, 2025):

Seems like this issue is related to our latest fix. May I check your installed version? Try the git log and also show it with pip list | grep ktrans. If older, maybe you can try reinstalling the latest version.

When Building 6617549, the following problem occurs: (venv) (base) user@user:/ktransformers$ export USE_NUMA=1 (venv) (base) user@user:/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation Using Python 3.11.11 environment at: venv × Failed to build ktransformers @ file:///home/user/ktransformers ├─▶ The build backend returned an error ╰─▶ Call to setuptools.build_meta.prepare_metadata_for_build_wheel failed (exit status: 1)

  [stderr]
  Traceback (most recent call last):
    File "<string>", line 14, in <module>
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel
      self.run_setup()
    File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup
      exec(code, locals())
    File "<string>", line 445, in <module>
  ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set.

  hint: This usually indicates a problem with the package or the build environment.
  1. Your CUDA_HOME may not set. You can echo $CUDA_HOME to see if it is correct and export the correct path.
  2. You may use the latest commit to check which is: 35ba63e
  3. But the quickest way to check is to download an older version of deepseek-ai/DeepSeekV3 config locally. Because the config file on the official net is updated, you can find an older one that fits in with our older version to see if this is the problem. And if it is, you can go ahead to use our latest version as it updates the new Deepseek v3 config support.

Thanks for the help! However, after creating the build, the following error occurred: Image

This may be related to some weird problem when installing with the previous cache. You can reinstall it without any cache by rm all the build folders in the current workspace. Note that the build folder also occurs in csrc where you need to clean too. And then reinstall it.

<!-- gh-comment-id:2799585753 --> @KMSorSMS commented on GitHub (Apr 13, 2025): > > > > Seems like this issue is related to our latest fix. May I check your installed version? Try the `git log` and also show it with `pip list | grep ktrans`. If older, maybe you can try reinstalling the latest version. > > > > > > > > > When Building [6617549](https://github.com/kvcache-ai/ktransformers/commit/6617549de0d1f40b0e9254105be7dcc11bf97de5), the following problem occurs: (venv) (base) user@user:~/ktransformers$ export USE_NUMA=1 (venv) (base) user@user:~/ktransformers$ KTRANSFORMERS_FORCE_BUILD=TRUE uv pip install . --no-build-isolation Using Python 3.11.11 environment at: venv × Failed to build `ktransformers @ file:///home/user/ktransformers` ├─▶ The build backend returned an error ╰─▶ Call to `setuptools.build_meta.prepare_metadata_for_build_wheel` failed (exit status: 1) > > > ``` > > > [stderr] > > > Traceback (most recent call last): > > > File "<string>", line 14, in <module> > > > File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 377, in prepare_metadata_for_build_wheel > > > self.run_setup() > > > File "/home/user/ktransformers/venv/lib/python3.11/site-packages/setuptools/build_meta.py", line 320, in run_setup > > > exec(code, locals()) > > > File "<string>", line 445, in <module> > > > ValueError: Unsupported backend: CUDA_HOME and MUSA_HOME are not set. > > > > > > hint: This usually indicates a problem with the package or the build environment. > > > ``` > > > > > > > > 1. Your CUDA_HOME may not set. You can echo $CUDA_HOME to see if it is correct and export the correct path. > > 2. You may use the latest commit to check which is: [35ba63e](https://github.com/kvcache-ai/ktransformers/tree/35ba63e259913ad16a1e49ad8a277235a43eb555) > > 3. But the quickest way to check is to download an older version of deepseek-ai/DeepSeekV3 config locally. Because the config file on the official net is updated, you can find an older one that fits in with our older version to see if this is the problem. And if it is, you can go ahead to use our latest version as it updates the new Deepseek v3 config support. > > Thanks for the help! However, after creating the build, the following error occurred: ![Image](https://github.com/user-attachments/assets/b250122e-bd4f-4fa5-baf4-570e2154b0fb) This may be related to some weird problem when installing with the previous cache. You can reinstall it without any cache by rm all the build folders in the current workspace. Note that the build folder also occurs in csrc where you need to clean too. And then reinstall it.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
kvcache-ai/ktransformers#775
No description provided.