ci : add short perplexity tests

Author: ggerganov

ci/run.sh

@@ -128,15 +128,20 @@ function gg_run_open_llama_3b_v2 {
     gg_wget models-mnt/open-llama/3B-v2/ https://huggingface.co/openlm-research/open_llama_3b_v2/resolve/main/pytorch_model.bin
     gg_wget models-mnt/open-llama/3B-v2/ https://huggingface.co/openlm-research/open_llama_3b_v2/raw/main/generation_config.json
 
+    gg_wget models-mnt/wikitext/ https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-raw-v1.zip
+    unzip -o models-mnt/wikitext/wikitext-2-raw-v1.zip -d models-mnt/wikitext/
+    head -n 60 models-mnt/wikitext/wikitext-2-raw/wiki.test.raw > models-mnt/wikitext/wikitext-2-raw/wiki.test-60.raw
+
+    path_models="../models-mnt/open-llama/3B-v2"
+    path_wiki="../models-mnt/wikitext/wikitext-2-raw"
+
     rm -rf build-ci-release && mkdir build-ci-release && cd build-ci-release
 
     set -e
 
     (time cmake -DCMAKE_BUILD_TYPE=Release -DLLAMA_QKK_64=1 .. ) 2>&1 | tee -a $OUT/${ci}-cmake.log
     (time make -j                                              ) 2>&1 | tee -a $OUT/${ci}-make.log
 
-    path_models="../models-mnt/open-llama/3B-v2"
-
     python3 ../convert.py ${path_models}
 
     model_f16="${path_models}/ggml-model-f16.bin"
@@ -150,6 +155,8 @@ function gg_run_open_llama_3b_v2 {
     model_q5_k="${path_models}/ggml-model-q5_k.bin"
     model_q6_k="${path_models}/ggml-model-q6_k.bin"
 
+    wiki_test_60="${path_wiki}/wiki.test-60.raw"
+
     ./bin/quantize ${model_f16} ${model_q8_0} q8_0
     ./bin/quantize ${model_f16} ${model_q4_0} q4_0
     ./bin/quantize ${model_f16} ${model_q4_1} q4_1
@@ -160,16 +167,27 @@ function gg_run_open_llama_3b_v2 {
     ./bin/quantize ${model_f16} ${model_q5_k} q5_k
     ./bin/quantize ${model_f16} ${model_q6_k} q6_k
 
-    (time ./bin/main --model ${model_f16}  -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-f16.log
-    (time ./bin/main --model ${model_q8_0} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
-    (time ./bin/main --model ${model_q4_0} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
-    (time ./bin/main --model ${model_q4_1} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
-    (time ./bin/main --model ${model_q5_0} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
-    (time ./bin/main --model ${model_q5_1} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
-    (time ./bin/main --model ${model_q3_k} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
-    (time ./bin/main --model ${model_q4_k} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
-    (time ./bin/main --model ${model_q5_k} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
-    (time ./bin/main --model ${model_q6_k} -s 1234 -n 64 -t 8 -p "I believe the meaning of life is") 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+    (time ./bin/main --model ${model_f16}  -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/main --model ${model_q8_0} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/main --model ${model_q4_0} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/main --model ${model_q4_1} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/main --model ${model_q5_0} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/main --model ${model_q5_1} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/main --model ${model_q3_k} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/main --model ${model_q4_k} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/main --model ${model_q5_k} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/main --model ${model_q6_k} -s 1234 -n 64 -p "I believe the meaning of life is" ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
+
+    (time ./bin/perplexity --model ${model_f16}  -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-f16.log
+    (time ./bin/perplexity --model ${model_q8_0} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q8_0.log
+    (time ./bin/perplexity --model ${model_q4_0} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_0.log
+    (time ./bin/perplexity --model ${model_q4_1} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_1.log
+    (time ./bin/perplexity --model ${model_q5_0} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_0.log
+    (time ./bin/perplexity --model ${model_q5_1} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_1.log
+    (time ./bin/perplexity --model ${model_q3_k} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q3_k.log
+    (time ./bin/perplexity --model ${model_q4_k} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q4_k.log
+    (time ./bin/perplexity --model ${model_q5_k} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q5_k.log
+    (time ./bin/perplexity --model ${model_q6_k} -n 64 -f ${wiki_test_60} -c 2048 ) 2>&1 | tee -a $OUT/${ci}-tg-q6_k.log
 
     set +e
 }