diff --git a/Makefile b/Makefile
index eba07670a..e0e5b96f1 100644
--- a/Makefile
+++ b/Makefile
@@ -235,7 +235,7 @@ extra.o: extra.cpp extra.h
 	$(CXX) $(CXXFLAGS) -c extra.cpp -o extra.o
 
 clean:
-	rm -vf *.o main quantize perplexity
+	rm -vf *.o main quantize perplexity embedding
 
 main: examples/main/main.cpp ggml.o llama.o common.o
 	$(CXX) $(CXXFLAGS) examples/main/main.cpp ggml.o llama.o common.o -o main $(LDFLAGS)
@@ -252,6 +252,9 @@ quantize: examples/quantize/quantize.cpp ggml.o llama.o
 perplexity: examples/perplexity/perplexity.cpp ggml.o llama.o common.o
 	$(CXX) $(CXXFLAGS) examples/perplexity/perplexity.cpp ggml.o llama.o common.o -o perplexity $(LDFLAGS)
 
+embedding: examples/embedding/embedding.cpp ggml.o llama.o common.o
+	$(CXX) $(CXXFLAGS) examples/embedding/embedding.cpp ggml.o llama.o common.o -o embedding $(LDFLAGS)
+
 #
 # Tests
 #
diff --git a/examples/embedding/CMakeLists.txt b/examples/embedding/CMakeLists.txt
index 88c425d4a..def5b831b 100644
--- a/examples/embedding/CMakeLists.txt
+++ b/examples/embedding/CMakeLists.txt
@@ -1,4 +1,4 @@
 set(TARGET embedding)
 add_executable(${TARGET} embedding.cpp)
-target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
+target_link_libraries(${TARGET} PRIVATE common llama ggml ${CMAKE_THREAD_LIBS_INIT})
 target_compile_features(${TARGET} PRIVATE cxx_std_11)
diff --git a/examples/main/CMakeLists.txt b/examples/main/CMakeLists.txt
index b2dcc2910..aa1f79406 100644
--- a/examples/main/CMakeLists.txt
+++ b/examples/main/CMakeLists.txt
@@ -1,4 +1,4 @@
 set(TARGET main)
 add_executable(${TARGET} main.cpp)
-target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
+target_link_libraries(${TARGET} PRIVATE common llama ggml ${CMAKE_THREAD_LIBS_INIT})
 target_compile_features(${TARGET} PRIVATE cxx_std_11)
diff --git a/examples/main/main.cpp b/examples/main/main.cpp
index e9478d541..66b7c2d5d 100644
--- a/examples/main/main.cpp
+++ b/examples/main/main.cpp
@@ -493,7 +493,7 @@ int main(int argc, char ** argv) {
         }
 
         // In interactive mode, respect the maximum number of tokens and drop back to user input when reached.
-        if (params.interactive && n_remain <= 0) {
+        if (params.interactive && n_remain <= 0 && params.n_predict != -1) {
             n_remain = params.n_predict;
             is_interacting = true;
         }
diff --git a/examples/perplexity/CMakeLists.txt b/examples/perplexity/CMakeLists.txt
index 5836df8b2..9bd8e376f 100644
--- a/examples/perplexity/CMakeLists.txt
+++ b/examples/perplexity/CMakeLists.txt
@@ -1,4 +1,4 @@
 set(TARGET perplexity)
 add_executable(${TARGET} perplexity.cpp)
-target_link_libraries(${TARGET} PRIVATE common llama ${CMAKE_THREAD_LIBS_INIT})
+target_link_libraries(${TARGET} PRIVATE common llama ggml ${CMAKE_THREAD_LIBS_INIT})
 target_compile_features(${TARGET} PRIVATE cxx_std_11)
diff --git a/examples/perplexity/perplexity.cpp b/examples/perplexity/perplexity.cpp
index f617ba365..75d526d3d 100644
--- a/examples/perplexity/perplexity.cpp
+++ b/examples/perplexity/perplexity.cpp
@@ -19,7 +19,7 @@ std::vector<double> softmax(const std::vector<float>& logits) {
 
 void perplexity(llama_context * ctx, const gpt_params & params) {
     // Download: https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-raw-v1.zip?ref=salesforce-research
-    // Run `./main --perplexity -m models/7B/ggml-model-q4_0.bin -f wiki.test.raw`
+    // Run `./perplexity -m models/7B/ggml-model-q4_0.bin -f wiki.test.raw`
     // Output: `perplexity: 13.5106 [114/114]`
     auto tokens = ::llama_tokenize(ctx, params.prompt, true);
 
diff --git a/examples/quantize/CMakeLists.txt b/examples/quantize/CMakeLists.txt
index fb27d4517..17a995bbd 100644
--- a/examples/quantize/CMakeLists.txt
+++ b/examples/quantize/CMakeLists.txt
@@ -1,4 +1,4 @@
 set(TARGET quantize)
 add_executable(${TARGET} quantize.cpp)
-target_link_libraries(${TARGET} PRIVATE llama ${CMAKE_THREAD_LIBS_INIT})
+target_link_libraries(${TARGET} PRIVATE llama ggml ${CMAKE_THREAD_LIBS_INIT})
 target_compile_features(${TARGET} PRIVATE cxx_std_11)
diff --git a/ggml.c b/ggml.c
index c9a4e8675..ec00e6317 100644
--- a/ggml.c
+++ b/ggml.c
@@ -1698,8 +1698,6 @@ inline static void ggml_vec_dot_q4_0(const int n, float * restrict s, const void
     // Horizontal sum of all lanes of the accumulator
     sumf = _mm512_reduce_add_ps( acc0 ) + _mm512_reduce_add_ps( acc1 );
 #elif defined(__AVX2__)
-    const size_t countBlocks = nb;
-
     // Initialize accumulator with zeros
     __m256 acc = _mm256_setzero_ps();
 
@@ -5806,23 +5804,28 @@ static void ggml_compute_forward_mul_mat_f32(
     const int ne02 = src0->ne[2];
     const int ne03 = src0->ne[3];
 
+#if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS)
     const int ne10 = src1->ne[0];
+#endif
     const int ne11 = src1->ne[1];
-    //const int ne12 = src1->ne[2];
-    //const int ne13 = src1->ne[3];
+#ifndef NDEBUG
+    const int ne12 = src1->ne[2];
+    const int ne13 = src1->ne[3];
 
-    //const int ne0  = dst->ne[0];
-    //const int ne1  = dst->ne[1];
-    //const int ne2  = dst->ne[2];
-    //const int ne3  = dst->ne[3];
-    //const int ne   = ne0*ne1*ne2*ne3;
+    const int ne0  = dst->ne[0];
+    const int ne1  = dst->ne[1];
+    const int ne2  = dst->ne[2];
+    const int ne3  = dst->ne[3];
 
-    //const int nb00 = src0->nb[0];
+    const int nb00 = src0->nb[0];
+#endif
     const int nb01 = src0->nb[1];
     const int nb02 = src0->nb[2];
     const int nb03 = src0->nb[3];
 
+#ifndef NDEBUG
     const int nb10 = src1->nb[0];
+#endif
     const int nb11 = src1->nb[1];
     const int nb12 = src1->nb[2];
     const int nb13 = src1->nb[3];
@@ -5840,8 +5843,9 @@ static void ggml_compute_forward_mul_mat_f32(
     assert(ne2  == ne12);
     assert(ne3  == ne13);
 
-    // TODO: we don't support permuted src0
+    // we don't support permuted src0 or src1
     assert(nb00 == sizeof(float));
+    assert(nb10 == sizeof(float));
 
     // dst cannot be transposed or permuted
     assert(nb0 == sizeof(float));
@@ -5859,8 +5863,6 @@ static void ggml_compute_forward_mul_mat_f32(
 
 #if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS)
     if (ggml_compute_forward_mul_mat_use_blas(src0, src1, dst)) {
-        GGML_ASSERT(nb10 == sizeof(float));
-
         if (params->ith != 0) {
             return;
         }
@@ -5903,9 +5905,6 @@ static void ggml_compute_forward_mul_mat_f32(
         return;
     }
 
-    // TODO: do not support transposed src1
-    assert(nb10 == sizeof(float));
-
     // parallelize by src0 rows using ggml_vec_dot_f32
 
     // total rows in src0
@@ -6169,7 +6168,6 @@ static void ggml_compute_forward_mul_mat_q4_0_f32(
     const int ne1  = dst->ne[1];
     const int ne2  = dst->ne[2];
     const int ne3  = dst->ne[3];
-    //const int ne   = ne0*ne1*ne2*ne3;
 
     const int nb00 = src0->nb[0];
     const int nb01 = src0->nb[1];
@@ -6194,8 +6192,9 @@ static void ggml_compute_forward_mul_mat_q4_0_f32(
     GGML_ASSERT(ne2  == ne12);
     GGML_ASSERT(ne3  == ne13);
 
-    // TODO: we don't support permuted src0
+    // we don't support permuted src0 or src1
     GGML_ASSERT(nb00 == (int) GGML_TYPE_SIZE[GGML_TYPE_Q4_0]);
+    GGML_ASSERT(nb10 == sizeof(float));
 
     // dst cannot be transposed or permuted
     GGML_ASSERT(nb0 == sizeof(float));
@@ -6213,8 +6212,6 @@ static void ggml_compute_forward_mul_mat_q4_0_f32(
 
 #if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS)
     if (ggml_compute_forward_mul_mat_use_blas(src0, src1, dst)) {
-        GGML_ASSERT(nb10 == sizeof(float));
-
         if (params->ith != 0) {
             return;
         }
@@ -6278,8 +6275,6 @@ static void ggml_compute_forward_mul_mat_q4_0_f32(
         return;
     }
 
-    // TODO: do not support transposed src1
-
     // parallelize by src0 rows using ggml_vec_dot_q4_0
 
     // total rows in src0
@@ -6354,7 +6349,6 @@ static void ggml_compute_forward_mul_mat_q4_1_f32(
     const int ne1  = dst->ne[1];
     const int ne2  = dst->ne[2];
     const int ne3  = dst->ne[3];
-    //const int ne   = ne0*ne1*ne2*ne3;
 
     const int nb00 = src0->nb[0];
     const int nb01 = src0->nb[1];
@@ -6379,8 +6373,9 @@ static void ggml_compute_forward_mul_mat_q4_1_f32(
     GGML_ASSERT(ne2  == ne12);
     GGML_ASSERT(ne3  == ne13);
 
-    // TODO: we don't support permuted src0
+    // we don't support permuted src0 or src1
     GGML_ASSERT(nb00 == (int) GGML_TYPE_SIZE[GGML_TYPE_Q4_1]);
+    GGML_ASSERT(nb10 == sizeof(float));
 
     // dst cannot be transposed or permuted
     GGML_ASSERT(nb0 == sizeof(float));
@@ -6398,8 +6393,6 @@ static void ggml_compute_forward_mul_mat_q4_1_f32(
 
 #if defined(GGML_USE_ACCELERATE) || defined(GGML_USE_OPENBLAS)
     if (ggml_compute_forward_mul_mat_use_blas(src0, src1, dst)) {
-        GGML_ASSERT(nb10 == sizeof(float));
-
         if (params->ith != 0) {
             return;
         }
@@ -6466,8 +6459,6 @@ static void ggml_compute_forward_mul_mat_q4_1_f32(
         return;
     }
 
-    // TODO: do not support transposed src1
-
     // parallelize by src0 rows using ggml_vec_dot_q4_1
 
     // total rows in src0
diff --git a/llamacpp.dll b/llamacpp.dll
index 11ecb4fa2..c5dc1bfff 100644
Binary files a/llamacpp.dll and b/llamacpp.dll differ
diff --git a/main.exe b/main.exe
index 6eeeeff11..3c5e6beb4 100644
Binary files a/main.exe and b/main.exe differ
diff --git a/quantize.exe b/quantize.exe
index 8ca9ddfbc..7f4f3f54b 100644
Binary files a/quantize.exe and b/quantize.exe differ