Add lint action (#32)

* temp

* fix lint

* csrc->src

* remove clang-format

* skip .rst

* skip doc

* clang-format

version

version

* mat_B

src/fastertransformer/models/llama/LlamaContextAttentionLayer.cc

@@ -199,7 +199,6 @@ inline void LlamaContextAttentionLayer<T>::forward(TensorMap*
                         quant_policy_,
                         weights->past_kv_scale.data());
 
-
     sync_check_cuda_error();
     if (use_fmha_) {
         fusedMultiHeadAttention(k_cache_ptrs,
@@ -226,8 +225,6 @@ inline void LlamaContextAttentionLayer<T>::forward(TensorMap*
                                   max_seq_len,
                                   quant_policy_,
                                   weights->past_kv_scale.data());
-
-        
     }
 
     //////////////////////////////////////////////
@@ -240,8 +237,6 @@ inline void LlamaContextAttentionLayer<T>::forward(TensorMap*
         sync_check_cuda_error();
     }
 
- 
-
     if (is_free_buffer_after_forward_ == true) {
         freeBuffer();
     }

src/fastertransformer/models/llama/LlamaContextAttentionLayer.h

@@ -16,7 +16,8 @@
  * limitations under the License.
  */
 
-// Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/attention_layers/GptContextAttentionLayer.h
+// Modified from
+// https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/attention_layers/GptContextAttentionLayer.h
 
 #pragma once
 

src/fastertransformer/models/llama/LlamaContextDecoder.h

@@ -15,7 +15,8 @@
  * limitations under the License.
  */
 
-// Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGptContextDecoder.h
+// Modified from
+// https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGptContextDecoder.h
 
 #pragma once
 

src/fastertransformer/models/llama/LlamaDecoder.h

@@ -16,8 +16,8 @@
  * limitations under the License.
  */
 
- // Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGptDecoder.h
-
+// Modified from
+// https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGptDecoder.h
 
 #include "src/fastertransformer/layers/BaseLayer.h"
 // #include "src/fastertransformer/layers/FfnLayer.h"
@@ -82,7 +82,6 @@ public:
                  bool             is_free_buffer_after_forward,
                  int              quant_policy),
 
-
         ~LlamaDecoder() override;
 
     virtual void forward(std::unordered_map<std::string, Tensor>*        output_tensors,

src/fastertransformer/models/llama/LlamaDecoderLayerWeight.cc

@@ -166,7 +166,8 @@ void LlamaDecoderLayerWeight<T>::loadModel(std::string dir_path, FtCudaDataType
     if (in.is_open()) {
         in.close();
         self_attn_weights.past_kv_scale = loadArrayFromBin({2}, scale_path);
-    } else {
+    }
+    else {
         self_attn_weights.past_kv_scale = {};
     }
 }

src/fastertransformer/models/llama/LlamaDecoderSelfAttentionLayer.h

@@ -15,7 +15,8 @@
  * limitations under the License.
  */
 
-// Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/attention_layers/DecoderSelfAttentionLayer.h
+// Modified from
+// https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/attention_layers/DecoderSelfAttentionLayer.h
 
 #pragma once
 

src/fastertransformer/models/llama/LlamaDenseWeight.h

@@ -15,7 +15,7 @@
  * limitations under the License.
  */
 
- // Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/DenseWeight.h
+// Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/DenseWeight.h
 
 #pragma once
 
@@ -25,8 +25,7 @@
 
 namespace fastertransformer {
 
-enum class WeightType : int
-{
+enum class WeightType : int {
     kFP32,
     kFP16,
     kFP8,  // not supported yet

src/fastertransformer/models/llama/LlamaFfnLayer.cc

@@ -15,7 +15,7 @@
  * limitations under the License.
  */
 
- // Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/FfnLayer.h
+// Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/FfnLayer.h
 
 #include "src/fastertransformer/models/llama/LlamaFfnLayer.h"
 #include "src/fastertransformer/kernels/activation_kernels.h"

src/fastertransformer/models/llama/LlamaFfnLayer.h

@@ -15,7 +15,7 @@
  * limitations under the License.
  */
 
- // Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/FfnLayer.cc
+// Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/layers/FfnLayer.cc
 
 #pragma once
 

src/fastertransformer/models/llama/LlamaV2.cc

@@ -17,7 +17,8 @@
  * limitations under the License.
  */
 
- // Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGpt.cc
+// Modified from
+// https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGpt.cc
 
 #include "src/fastertransformer/models/llama/LlamaV2.h"
 #include "src/fastertransformer/kernels/decoding_kernels.h"
@@ -97,7 +98,8 @@ LlamaV2<T>::LlamaV2(size_t                       head_num,
             FT_LOG_ERROR("use_context_fmha not support int8");
             assert(0);
         }
-    } else {
+    }
+    else {
         elem_bits = sizeof(T) * 8;
     }
     kv_cache_mgr_ = std::make_unique<LlamaCacheManager>(num_layer_,
@@ -213,7 +215,7 @@ void LlamaV2<T>::contextDecode(T*         deocder_output,
                                              static_cast<T*>(nullptr),
                                              pPromptTuningParam<T>{},
                                              input_ids,
-                                             0,  // only used for postion encoding
+                                             0,  // only used for position encoding
                                              token_num,
                                              token_num,
                                              1,

src/fastertransformer/models/llama/LlamaV2.h

@@ -16,7 +16,8 @@
  * limitations under the License.
  */
 
-// Modified from https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGpt.h
+// Modified from
+// https://github.com/NVIDIA/FasterTransformer/blob/main/src/fastertransformer/models/multi_gpu_gpt/ParallelGpt.h
 
 #pragma once