GitLab

update llama4 tutorial

update llama4 tutorial

balance_serve: Add g++ to compiler list

In some OS distributions, g++ exists in the following form:

  # ls -l /usr/bin/g++*
  -rwxr-xr-x 4 root root 985784 Dec  9 12:51 /usr/bin/g++

So make sure to add g++ to the compiler list as well.

Update attention.py

Update balance-serve.md

format kvc2, delete quant_configs, move model_configs to ~/.ktransfor…

fix compile, add abi check to setup.py

Update balance-serve.md

Fix TypeError when invoke KLinearCPUInfer.forward()

fix temperature=0, flashinfer sample error

Fix compile

Fix the following error:

  File "/home/aubrey/work/ktransformers/ktransformers/operators/linear.py", line 825, in forward
    y = self.generate_linear.forward(x, bsz_tensor)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: KLinearCPUInfer.forward() takes 2 positional arguments but 3 were given

Update balance-serve.md

📝 Docs: Clarify CMake version requirement for CUDA dialects

Adds a note explaining that default CMake versions on systems like
Ubuntu 22.04 LTS might not support newer CUDA dialects (e.g., CUDA 20),
leading to specific build errors.

Recommends installing a newer CMake via the Kitware APT repository
as a resolution. This helps users troubleshoot errors like:
"Target ... requires the language dialect 'CUDA20', but CMake does not know the compile flags..."

🔖 release v0.2.4post1

Fix bug with non-base-multiple chunk_size, update test examples, and …

Fix bug with non-base-multiple chunk_size, update test examples, and resolve issue with writing model_config. Hugging Face URL input is still unsupported.

🔧 update config.yaml setting default config

Update modeling_deepseek_v3.py

slove [Bug] #1023

文档更新：model_path名字要求以及在示例中添加force_think

* add documentations about `--model_path` requirements
* add `--force_think` in doc (most users would run R1 and would want it to provide reasoning process)

Only modified the mixed single and double quotes in server/config/config.py