fix build

Author: pavel-esir

.github/workflows/causal_lm_cpp.yml

@@ -32,12 +32,12 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model openlm-research/open_llama_3b_v2 open_llama_3b_v2
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: greedy_causal_lm
         run: |
           source ./ov/setupvars.sh
-          ./build/greedy_causal_lm ./open_llama_3b_v2/ "return 0"
+          ./build/text_generation/causal_lm/cpp/greedy_causal_lm ./open_llama_3b_v2/ "return 0"
 
   cpp-beam_search_causal_lm-ubuntu:
     runs-on: ubuntu-20.04
@@ -60,13 +60,13 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model TinyLlama/TinyLlama-1.1B-Chat-v1.0 TinyLlama-1.1B-Chat-v1.0
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: Compare
         run: |
           source ./ov/setupvars.sh
 
-          timeout 25s ./build/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "Why is the Sun yellow?" > ./pred.txt
+          timeout 25s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "Why is the Sun yellow?" > ./pred.txt
           python -c "
           import transformers
           with open('pred.txt', 'r') as file:
@@ -82,7 +82,7 @@ jobs:
           "
           echo "Why is the Sun yellow?" passed
 
-          timeout 25s ./build/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ 69 > ./pred.txt
+          timeout 25s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ 69 > ./pred.txt
           python -c "
           import transformers
           with open('pred.txt', 'r') as file:
@@ -98,7 +98,7 @@ jobs:
           "
           echo "69" passed
 
-          timeout 25s ./build/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ Hi > ./pred.txt
+          timeout 25s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ Hi > ./pred.txt
           python -c "
           import transformers
           with open('pred.txt', 'r') as file:
@@ -114,7 +114,7 @@ jobs:
           "
           echo "Hi" passed
 
-          timeout 25s ./build/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "return 0" > ./pred.txt
+          timeout 25s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "return 0" > ./pred.txt
           python -c "
           import transformers
           with open('pred.txt', 'r') as file:
@@ -130,7 +130,7 @@ jobs:
           "
           echo "return 0" passed
 
-          ./build/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "你好！ 你好嗎？" > ./pred.txt
+          ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "你好！ 你好嗎？" > ./pred.txt
           python -c "
           import transformers
           with open('pred.txt', 'r') as file:
@@ -146,7 +146,7 @@ jobs:
           "
           echo "你好！ 你好嗎？" passed
 
-          timeout 1m ./build/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "Alan Turing was a" "return 0" "你好！ 你好嗎？" > ./pred.txt
+          timeout 1m ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "Alan Turing was a" "return 0" "你好！ 你好嗎？" > ./pred.txt
           python -c "
           import transformers
           with open('pred.txt', 'r') as file:
@@ -188,7 +188,7 @@ jobs:
           python -m pip install --upgrade-strategy eager -r text_generation/causal_lm/cpp/requirements.txt
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model TinyLlama/TinyLlama-1.1B-Chat-v1.0 TinyLlama-1.1B-Chat-v1.0
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: Compare
         shell: cmd
@@ -229,12 +229,12 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model Qwen/Qwen-7B-Chat Qwen-7B-Chat
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: Compare
         run: |
           source ./ov/setupvars.sh
-          timeout 50s ./build/beam_search_causal_lm ./Qwen-7B-Chat/ 69 > ./pred.txt
+          timeout 50s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./Qwen-7B-Chat/ 69 > ./pred.txt
 
   cpp-beam_search_causal_lm-Qwen1_5-7B-Chat:
     runs-on: ubuntu-20.04-16-cores
@@ -257,12 +257,12 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model Qwen/Qwen1.5-7B-Chat Qwen1.5-7B-Chat
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: Run
         run: |
           source ./ov/setupvars.sh
-          timeout 50s ./build/beam_search_causal_lm ./Qwen1.5-7B-Chat/ "你好！" > ./pred_qwen15.txt
+          timeout 50s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./Qwen1.5-7B-Chat/ "你好！" > ./pred_qwen15.txt
 
   cpp-beam_search_causal_lm-Phi-2:
     runs-on: ubuntu-20.04-16-cores
@@ -285,12 +285,12 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model microsoft/phi-2 phi-2
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j 15
       - name: Compare
         run: |
           source ./ov/setupvars.sh
-          timeout 50s ./build/beam_search_causal_lm ./phi-2/ 69 > ./pred.txt
+          timeout 50s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./phi-2/ 69 > ./pred.txt
 
   cpp-beam_search_causal_lm-notus-7b-v1:
     runs-on: ubuntu-20.04-16-cores
@@ -313,12 +313,12 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model argilla/notus-7b-v1 notus-7b-v1
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: Compare
         run: |
           source ./ov/setupvars.sh
-          timeout 50s ./build/beam_search_causal_lm ./notus-7b-v1/ 69 > ./pred.txt
+          timeout 50s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./notus-7b-v1/ 69 > ./pred.txt
 
   cpp-speculative_decoding_lm-ubuntu:
     runs-on: ubuntu-20.04-16-cores
@@ -342,13 +342,13 @@ jobs:
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model databricks/dolly-v2-3b dolly-v2-3b
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model databricks/dolly-v2-7b dolly-v2-7b
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: run and compare
         run: |
           source ./ov/setupvars.sh
           ./build/speculative_decoding_lm ./dolly-v2-3b/ ./dolly-v2-7b/ "Alan Turing was a" > predictions_speculative.txt
-          ./build/greedy_causal_lm ./dolly-v2-7b/ "Alan Turing was a" > predictions_greedy.txt
+          ./build/text_generation/causal_lm/cpp/greedy_causal_lm ./dolly-v2-7b/ "Alan Turing was a" > predictions_greedy.txt
           python -c "
           with open('predictions_greedy.txt', 'r') as f:
               predicted_greedy = f.readline()
@@ -380,7 +380,7 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model TinyLlama/TinyLlama-1.1B-Chat-v1.0 TinyLlama-1.1B-Chat-v1.0
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - name: run and compare
         run: |
@@ -394,7 +394,7 @@ jobs:
           A:' > ./prompt.txt
 
           ./build/prompt_lookup_decoding_lm ./TinyLlama-1.1B-Chat-v1.0/ "$(<prompt.txt)" > predictions_prompt_lookup.txt
-          ./build/greedy_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "$(<prompt.txt)" > predictions_greedy.txt
+          ./build/text_generation/causal_lm/cpp/greedy_causal_lm ./TinyLlama-1.1B-Chat-v1.0/ "$(<prompt.txt)" > predictions_greedy.txt
           python -c "
           with open('predictions_greedy.txt', 'r') as f:
               predicted_greedy = f.readline()
@@ -425,13 +425,13 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model microsoft/phi-1_5 phi-1_5
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j 15
       - name: Run Generation
         run: |
           source ./ov/setupvars.sh
-          timeout 50s ./build/greedy_causal_lm ./phi-1_5/ "Alan Turing was a" > ./pred_greedy.txt
-          timeout 50s ./build/beam_search_causal_lm ./phi-1_5/ "Alan Turing was a" > ./pred_beam.txt
+          timeout 50s ./build/text_generation/causal_lm/cpp/greedy_causal_lm ./phi-1_5/ "Alan Turing was a" > ./pred_greedy.txt
+          timeout 50s ./build/text_generation/causal_lm/cpp/beam_search_causal_lm ./phi-1_5/ "Alan Turing was a" > ./pred_beam.txt
       - name: Compare
         run: |
           python -c "
@@ -470,13 +470,13 @@ jobs:
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers]
           sudo apt-get install libtbb-dev
           optimum-cli export openvino --trust-remote-code --weight-format fp16 --model ikala/redpajama-3b-chat redpajama-3b-chat
-          cmake -DCMAKE_BUILD_TYPE=Release -S ./text_generation/causal_lm/cpp/ -B ./build/
+          cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release -j
       - run: source ./ov/setupvars.sh && convert_tokenizer ./redpajama-3b-chat/ --output ./redpajama-3b-chat/ --with-detokenizer --trust-remote-code
       - name: Run Generation
         run: |
           source ./ov/setupvars.sh
-          timeout 50s ./build/greedy_causal_lm ./redpajama-3b-chat/ "Alan Turing was a" > ./pred_greedy.txt
+          timeout 50s ./build/text_generation/causal_lm/cpp/greedy_causal_lm ./redpajama-3b-chat/ "Alan Turing was a" > ./pred_greedy.txt
       - name: Compare
         run: |
           python -c "