[refactor]: Change named 'KT-SFT' to 'kt-sft' (#1626)

* Change named 'KT-SFT' to 'kt-sft'

* [docs]: update kt-sft name

---------

Co-authored-by: ZiWei Yuan <yzwliam@126.com>

README.md

@@ -8,12 +8,12 @@
 
 </p>
   <h3>A Flexible Framework for Experiencing Cutting-edge LLM Inference/Fine-tune Optimizations</h3>
-  <strong><a href="#-overview">🎯 Overview</a> | <a href="#-kt-kernel---high-performance-inference-kernels">🚀 kt-kernel</a> | <a href="#-kt-sft---fine-tuning-framework">🎓 KT-SFT</a> | <a href="#-citation">🔥 Citation</a> | <a href="https://github.com/kvcache-ai/ktransformers/issues/1582">🚀 Roadmap(2025Q4)</a>  </strong>
+  <strong><a href="#-overview">🎯 Overview</a> | <a href="#-kt-kernel---high-performance-inference-kernels">🚀 kt-kernel</a> | <a href="#-kt-sft---fine-tuning-framework">🎓 kt-sft</a> | <a href="#-citation">🔥 Citation</a> | <a href="https://github.com/kvcache-ai/ktransformers/issues/1582">🚀 Roadmap(2025Q4)</a>  </strong>
 </div>
 
 ## 🎯 Overview
 
-KTransformers is a research project focused on efficient inference and fine-tuning of large language models through CPU-GPU heterogeneous computing. The project has evolved into **two core modules**: [kt-kernel](./kt-kernel/) and [KT-SFT](./KT-SFT/).
+KTransformers is a research project focused on efficient inference and fine-tuning of large language models through CPU-GPU heterogeneous computing. The project has evolved into **two core modules**: [kt-kernel](./kt-kernel/) and [kt-sft](./kt-sft/).
 
 ## 🔥 Updates
 
@@ -79,7 +79,7 @@ pip install .
 
 ---
 
-### 🎓 [KT-SFT](./KT-SFT/) - Fine-Tuning Framework
+### 🎓 [kt-sft](./kt-sft/) - Fine-Tuning Framework
 
 KTransformers × LLaMA-Factory integration for ultra-large MoE model fine-tuning.
 
@@ -101,12 +101,12 @@ KTransformers × LLaMA-Factory integration for ultra-large MoE model fine-tuning
 
 **Quick Start:**
 ```bash
-cd KT-SFT
-# Install environment following KT-SFT/README.md
+cd kt-sft
+# Install environment following kt-sft/README.md
 USE_KT=1 llamafactory-cli train examples/train_lora/deepseek3_lora_sft_kt.yaml
 ```
 
-👉 **[Full Documentation →](./KT-SFT/README.md)**
+👉 **[Full Documentation →](./kt-sft/README.md)**
 
 ---
 

README_ZH.md

@@ -8,12 +8,12 @@
 
 </p>
   <h3>一个用于体验尖端 LLM 推理/微调优化的灵活框架</h3>
-  <strong><a href="#-概览">🎯 概览</a> | <a href="#-kt-kernel---高性能推理内核">🚀 kt-kernel</a> | <a href="#-kt-sft---微调框架">🎓 KT-SFT</a> | <a href="#-引用">🔥 引用</a> </strong>
+  <strong><a href="#-概览">🎯 概览</a> | <a href="#-kt-kernel---高性能推理内核">🚀 kt-kernel</a> | <a href="#-kt-sft---微调框架">🎓 kt-sft</a> | <a href="#-引用">🔥 引用</a> </strong>
 </div>
 
 ## 🎯 概览
 
-KTransformers 是一个专注于通过 CPU-GPU 异构计算实现大语言模型高效推理和微调的研究项目。该项目已发展为**两个核心模块**：[kt-kernel](./kt-kernel/) 和 [KT-SFT](./KT-SFT/)。
+KTransformers 是一个专注于通过 CPU-GPU 异构计算实现大语言模型高效推理和微调的研究项目。该项目已发展为**两个核心模块**：[kt-kernel](./kt-kernel/) 和 [kt-sft](./kt-sft/)。
 
 ## 🔥 更新
 
@@ -78,7 +78,7 @@ pip install .
 
 ---
 
-### 🎓 [KT-SFT](./KT-SFT/) - 微调框架
+### 🎓 [kt-sft](./kt-sft/) - 微调框架
 
 KTransformers × LLaMA-Factory 集成，用于超大型 MoE 模型微调。
 
@@ -100,12 +100,12 @@ KTransformers × LLaMA-Factory 集成，用于超大型 MoE 模型微调。
 
 **快速开始：**
 ```bash
-cd KT-SFT
-# 按照 KT-SFT/README.md 安装环境
+cd kt-sft
+# 按照 kt-sft/README.md 安装环境
 USE_KT=1 llamafactory-cli train examples/train_lora/deepseek3_lora_sft_kt.yaml
 ```
 
-👉 **[完整文档 →](./KT-SFT/README.md)**
+👉 **[完整文档 →](./kt-sft/README.md)**
 
 ---
 

archive/README.md

@@ -9,7 +9,7 @@
 
 ## 🎯 Overview
 
-KTransformers is a research project focused on efficient inference and fine-tuning of large language models through CPU-GPU heterogeneous computing. The project has evolved into **two core modules**: [kt-kernel](./kt-kernel/) and [KT-SFT](./KT-SFT/).
+KTransformers is a research project focused on efficient inference and fine-tuning of large language models through CPU-GPU heterogeneous computing. The project has evolved into **two core modules**: [kt-kernel](./kt-kernel/) and [kt-sft](./kt-sft/).
 
 ## 🔥 Updates
 
@@ -67,7 +67,7 @@ pip install .
 
 ---
 
-### 🎓 [KT-SFT](./KT-SFT/) - Fine-Tuning Framework
+### 🎓 [kt-sft](./kt-sft/) - Fine-Tuning Framework
 
 KTransformers × LLaMA-Factory integration for ultra-large MoE model fine-tuning.
 
@@ -89,12 +89,12 @@ KTransformers × LLaMA-Factory integration for ultra-large MoE model fine-tuning
 
 **Quick Start:**
 ```bash
-cd KT-SFT
-# Install environment following KT-SFT/README.md
+cd kt-sft
+# Install environment following kt-sft/README.md
 USE_KT=1 llamafactory-cli train examples/train_lora/deepseek3_lora_sft_kt.yaml
 ```
 
-👉 **[Full Documentation →](./KT-SFT/README.md)**
+👉 **[Full Documentation →](./kt-sft/README.md)**
 
 ---
 

archive/README_ZH.md

@@ -9,7 +9,7 @@
 
 ## 🎯 项目概述
 
-KTransformers 是一个专注于大语言模型高效推理和微调的研究项目，通过 CPU-GPU 异构计算实现资源受限环境下的模型部署。项目已演进为**两个核心模块**：[kt-kernel](./kt-kernel/) 和 [KT-SFT](./KT-SFT/)。
+KTransformers 是一个专注于大语言模型高效推理和微调的研究项目，通过 CPU-GPU 异构计算实现资源受限环境下的模型部署。项目已演进为**两个核心模块**：[kt-kernel](./kt-kernel/) 和 [kt-sft](./kt-sft/)。
 
 ## 🔥 更新
 
@@ -66,7 +66,7 @@ pip install .
 
 ---
 
-### 🎓 [KT-SFT](./KT-SFT/) - 微调框架
+### 🎓 [kt-sft](./kt-sft/) - 微调框架
 
 KTransformers × LLaMA-Factory 集成，支持超大 MoE 模型微调。
 
@@ -86,12 +86,12 @@ KTransformers × LLaMA-Factory 集成，支持超大 MoE 模型微调。
 
 **快速开始：**
 ```bash
-cd KT-SFT
-# 按照 KT-SFT/README.md 安装环境
+cd kt-sft
+# 按照 kt-sft/README.md 安装环境
 USE_KT=1 llamafactory-cli train examples/train_lora/deepseek3_lora_sft_kt.yaml
 ```
 
-👉 **[完整文档 →](./KT-SFT/README.md)**
+👉 **[完整文档 →](./kt-sft/README.md)**
 
 ---
 

doc/SUMMARY.md

@@ -3,11 +3,11 @@
 [Introduction](./README.md)
 # Install & Usage
 - [For kt-kernel](en/kt-kernel/kt-kernel_intro.md)
-- [For SFT](en/SFT/KTransformers-Fine-Tuning_User-Guide.md)
+- [For kt-sft](en/SFT/KTransformers-Fine-Tuning_User-Guide.md)
 
 # Tutorial 
-- [SFT part](en/SFT/README.md)
-  - [SFT developer tech notes](en/SFT/KTransformers-Fine-Tuning_Developer-Technical-Notes.md)
+- [kt-sft part](en/SFT/README.md)
+  - [kt-sft developer tech notes](en/SFT/KTransformers-Fine-Tuning_Developer-Technical-Notes.md)
   - [Injection Tutorial](en/SFT/injection_tutorial.md)
   <!-- - [Multi-GPU Tutorial](en/multi-gpu-tutorial.md) -->
   <!-- - [Use FP8 GPU Kernel](en/fp8_kernel.md) -->

doc/en/SFT/README.md

@@ -1 +1 @@
-# SFT Docs
+# kt-sft Docs

kt-sft/csrc/ktransformers_ext/examples/test_sft_amx_moe.py

@@ -35,7 +35,7 @@ gradtype = torch.bfloat16
 # torch.backends.cuda.matmul.allow_tf32 = False
 
 import shutil
-folder_path = "/home/lpl/KT-SFT/debug"
+folder_path = "/home/lpl/kt-sft/debug"
 if os.path.exists(folder_path):
     shutil.rmtree(folder_path)
 os.makedirs(folder_path)
@@ -650,13 +650,13 @@ def manual_check(experts_ids):
         
         down_ba_ori = get_tensor(f"cpp_layer0_E_End{experts_idx}_down_ba_ori_", (expert_token_counts[experts_idx], intermediate_size))
 
-        # with open(f"/home/lpl/KT-SFT/debug/cpp_{experts_idx}_down_ba_ori_view.txt", "w") as f:
+        # with open(f"/home/lpl/kt-sft/debug/cpp_{experts_idx}_down_ba_ori_view.txt", "w") as f:
         #     f.write(str(down_ba_ori))   
         
     
         down_output_grad = get_tensor(f"cpp_layer0_E_End{experts_idx}_down_output_grad_", (expert_token_counts[experts_idx], hidden_size))
 
-        # with open(f"/home/lpl/KT-SFT/debug/cpp_{experts_idx}_down_t_ba_ori_view.txt", "w") as f:
+        # with open(f"/home/lpl/kt-sft/debug/cpp_{experts_idx}_down_t_ba_ori_view.txt", "w") as f:
         #     f.write(str(down_output_grad))
             
         
@@ -674,10 +674,10 @@ def manual_check(experts_ids):
         py_down_t_ba = torch.load(f"debug/py_layer0_E_End{experts_idx}_down_output_grad_.pt")
         py_down_ba = torch.load(f"debug/py_layer0_E_End{experts_idx}_gate_output_.pt")
 
-        # with open(f"/home/lpl/KT-SFT/debug/py_{experts_idx}_down_t_ba_ori_view.txt", "w") as f:
+        # with open(f"/home/lpl/kt-sft/debug/py_{experts_idx}_down_t_ba_ori_view.txt", "w") as f:
         #     f.write(str(py_down_t_ba))
         
-        # with open(f"/home/lpl/KT-SFT/debug/py_{experts_idx}_down_ba_ori_view.txt", "w") as f:
+        # with open(f"/home/lpl/kt-sft/debug/py_{experts_idx}_down_ba_ori_view.txt", "w") as f:
         #     f.write(str(py_down_ba))
             
         print(f"cpp_{experts_idx}_down_ba_ori_:{down_ba_ori}")