[CPP Graph] Support StarCoder-15.5B inference (#1158)

intel_extension_for_transformers/backends/neural_engine/graph/README.md

@@ -43,6 +43,10 @@ python scripts/convert_falcon.py ${input_model_name_or_path} --outtype f32 --out
 
 ./build/bin/quant_falcon --model_file ${output_path}/ne-f32.bin --out_file ${output_path}/ne-q4_j.bin --bits 4
 
+# convert the pytorch starcoder model to llama.cpp format
+python scripts/convert_starcoder.py --model={input_model_name_or_path} --outfile={output_path}/ne-fp32.bin --outtype=fp32
+
+./build/bin/quant_starcoder --model_file ${output_path}/ne-f32.bin --out_file ${output_path}/ne-q4_j.bin --bits 4
 ```
 
 ### Run Models
@@ -52,7 +56,8 @@ Running LLAMA model, for details please refer to [LLaMA model documentation](./a
 OMP_NUM_THREADS=56 numactl -m 0 -C 0-55 ./build/bin/main_llama -m ~/llama.cpp/models/ne-model-q4_j.bin --seed 12 -c 512 -b 1024 -n 256 --keep 48 -t 56 --repeat-penalty 1.0 --color -p "She opened the door and see"
 ```
 
-Running GPT-NEOX/ MPT / FALCON / GPT-J model, please use `main_gptneox` / `main_mpt` / `main_falcon` / `main_gptj`.
+Running GPT-NEOX / MPT / FALCON / / GPT-J / STARCODER model, please use `main_gptneox` / `main_mpt` / `main_falcon` / `main_starcoder` (Please type **prompt about codes** when use `STARCODER`. For example, `-p "def fibonnaci("`).
+
 
 ```bash
 OMP_NUM_THREADS=56 numactl -m 0 -C 0-55 ./build/bin/main_gptneox -m ${output_path}/ne-q8.bin --seed 12 -c 512 -b 1024 -n 256 -t 56 --repeat-penalty 1.0 -p "She opened the door and see"
@@ -67,4 +72,4 @@ python gptj_binding.py
 ```
 
 ### Supported model
-Now we supports [GPT-NeoX](https://github.com/EleutherAI/gpt-neox), [LLaMA](https://github.com/facebookresearch/llama), [MPT](https://huggingface.co/mosaicml/mpt-7b), [FALCON](https://huggingface.co/tiiuae/falcon-7b), [GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj)
+Now we supports [GPT-NeoX](https://github.com/EleutherAI/gpt-neox), [LLaMA](https://github.com/facebookresearch/llama), [MPT](https://huggingface.co/mosaicml/mpt-7b), [FALCON](https://huggingface.co/tiiuae/falcon-7b), [STARCODER](https://huggingface.co/bigcode/starcoder), [GPT-J](https://huggingface.co/docs/transformers/model_doc/gptj).

intel_extension_for_transformers/backends/neural_engine/graph/application/CMakeLists.txt

@@ -31,4 +31,5 @@ add_subdirectory(ChatLLAMA)
 add_subdirectory(ChatGPTNEOX)
 add_subdirectory(ChatMPT)
 add_subdirectory(ChatFALCON)
+add_subdirectory(ChatSTARCODER)
 add_subdirectory(ChatGPTJ)

intel_extension_for_transformers/backends/neural_engine/graph/application/ChatSTARCODER/CMakeLists.txt

@@ -0,0 +1,29 @@
+#  Copyright (c) 2023 Intel Corporation
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+set(TARGET main_starcoder)
+add_executable_w_warning(${TARGET} main_starcoder.cpp)
+target_link_libraries(${TARGET} PUBLIC ne_layers common ${CMAKE_THREAD_LIBS_INIT})
+target_compile_features(${TARGET} PRIVATE cxx_std_11)
+if(TARGET BUILD_INFO)
+  add_dependencies(${TARGET} BUILD_INFO)
+endif()
+
+set(TARGET quant_starcoder)
+add_executable_w_warning(${TARGET} quant_starcoder.cpp)
+target_link_libraries(${TARGET} PUBLIC ne_layers common ${CMAKE_THREAD_LIBS_INIT})
+target_compile_features(${TARGET} PRIVATE cxx_std_11)
+if(TARGET BUILD_INFO)
+  add_dependencies(${TARGET} BUILD_INFO)
+endif()