diff --git a/.github/workflows/causal_lm_cpp.yml b/.github/workflows/causal_lm_cpp.yml
index 2263277b68..355a8d8466 100644
--- a/.github/workflows/causal_lm_cpp.yml
+++ b/.github/workflows/causal_lm_cpp.yml
@@ -376,7 +376,7 @@ jobs:
         run: |
           source ./ov/setupvars.sh
           ./build/samples/cpp/speculative_decoding_lm/speculative_decoding_lm ./dolly-v2-3b/ ./dolly-v2-7b/ "Alan Turing was a" > predictions_speculative.txt
-          ./build/samples/cpp/speculative_decoding_lm/speculative_decoding_lm ./dolly-v2-3b/ ./dolly-v2-7b/ "Alan Turing was a" > predictions_greedy.txt
+          ./build/samples/cpp/greedy_causal_lm/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()
@@ -420,7 +420,7 @@ jobs:
           A:' > ./prompt.txt
 
           ./build/samples/cpp/prompt_lookup_decoding_lm/prompt_lookup_decoding_lm ./TinyLlama-1.1B-Chat-v1.0/ "$(<prompt.txt)" > predictions_prompt_lookup.txt
-          ./build/samples/cpp/prompt_lookup_decoding_lm/prompt_lookup_decoding_lm ./TinyLlama-1.1B-Chat-v1.0/ "$(<prompt.txt)" > predictions_greedy.txt
+          ./build/samples/cpp/greedy_causal_lm/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()
diff --git a/samples/cpp/prompt_lookup_decoding_lm/prompt_lookup_decoding_lm.cpp b/samples/cpp/prompt_lookup_decoding_lm/prompt_lookup_decoding_lm.cpp
index 51ac654aac..dc70d25c2c 100644
--- a/samples/cpp/prompt_lookup_decoding_lm/prompt_lookup_decoding_lm.cpp
+++ b/samples/cpp/prompt_lookup_decoding_lm/prompt_lookup_decoding_lm.cpp
@@ -1,9 +1,9 @@
 // Copyright (C) 2023-2024 Intel Corporation
 // SPDX-License-Identifier: Apache-2.0
 
-#include <string_view>
 #include <openvino/core/parallel.hpp>
 #include <openvino/openvino.hpp>
+#include <string_view>
 
 namespace {
 
@@ -58,7 +58,7 @@ struct TextStreamer {
     void end() {
         std::string text = detokenize(detokenizer, token_cache);
         if (text.size() <= print_len)
-            return ;
+            return;
         std::cout << std::string_view{text.data() + print_len, text.size() - print_len} << '\n';
         token_cache.clear();
         print_len = 0;
@@ -75,10 +75,7 @@ ov::Tensor trimm_tensor(ov::Tensor& tensor, uint64_t seq_len_axis, uint64_t new_
 
     auto old_tensor_data = tensor.data<float>();
     auto shape = tensor.get_shape();
-    size_t batch_size = shape[0];
-    size_t num_kv_heads = shape[1];
-    size_t old_seq_len = shape[2];
-    size_t head_size = shape[3];
+    size_t old_seq_len = shape[seq_len_axis];
 
     OPENVINO_ASSERT(new_seq_len <= old_seq_len);
 
@@ -86,14 +83,16 @@ ov::Tensor trimm_tensor(ov::Tensor& tensor, uint64_t seq_len_axis, uint64_t new_
     if (old_seq_len == new_seq_len)
         return tensor;
 
+    shape[seq_len_axis] = new_seq_len;
+
     if (seq_len_axis == 0) {
-        shape[0] = new_seq_len;
         tensor.set_shape(shape);
         return tensor;
     }
 
     ov::Coordinate new_shape_begin{0, 0, 0, 0};
-    ov::Coordinate new_shape_end{batch_size, num_kv_heads, new_seq_len, head_size};
+    ov::Coordinate new_shape_end{shape};
+
     auto new_tensor = ov::Tensor(tensor, new_shape_begin, new_shape_end);
 
     return new_tensor;
diff --git a/samples/cpp/speculative_decoding_lm/speculative_decoding_lm.cpp b/samples/cpp/speculative_decoding_lm/speculative_decoding_lm.cpp
index 4927b7d795..ba610574e8 100644
--- a/samples/cpp/speculative_decoding_lm/speculative_decoding_lm.cpp
+++ b/samples/cpp/speculative_decoding_lm/speculative_decoding_lm.cpp
@@ -6,13 +6,13 @@
 #include <openvino/openvino.hpp>
 #include <random>
 
-constexpr size_t BATCH_SIZE = 1;
+namespace {
 
+constexpr size_t BATCH_SIZE = 1;
 // sequence length axis in key/values tensors, for most cases [BATCH_SIZE, num_kv_heads, seq_len, head_size],
 // threfore usually SEQ_LEN_AXIS = 2
 constexpr size_t SEQ_LEN_AXIS = 2;
 
-namespace {
 std::pair<ov::Tensor, ov::Tensor> tokenize(ov::InferRequest& tokenizer, std::string&& prompt) {
     tokenizer.set_input_tensor(ov::Tensor{ov::element::string, {BATCH_SIZE}, &prompt});
     tokenizer.infer();
@@ -58,7 +58,7 @@ struct TextStreamer {
     void end() {
         std::string text = detokenize(detokenizer, token_cache);
         if (text.size() <= print_len)
-            return ;
+            return;
         std::cout << std::string_view{text.data() + print_len, text.size() - print_len} << '\n';
         token_cache.clear();
         print_len = 0;
@@ -75,10 +75,7 @@ ov::Tensor trimm_tensor(ov::Tensor& tensor, uint64_t seq_len_axis, uint64_t new_
 
     auto old_tensor_data = tensor.data<float>();
     auto shape = tensor.get_shape();
-    size_t batch_size = shape[0];
-    size_t num_kv_heads = shape[1];
-    size_t old_seq_len = shape[2];
-    size_t head_size = shape[3];
+    size_t old_seq_len = shape[seq_len_axis];
 
     OPENVINO_ASSERT(new_seq_len <= old_seq_len);
 
@@ -86,14 +83,16 @@ ov::Tensor trimm_tensor(ov::Tensor& tensor, uint64_t seq_len_axis, uint64_t new_
     if (old_seq_len == new_seq_len)
         return tensor;
 
+    shape[seq_len_axis] = new_seq_len;
+
     if (seq_len_axis == 0) {
-        shape[0] = new_seq_len;
         tensor.set_shape(shape);
         return tensor;
     }
 
     ov::Coordinate new_shape_begin{0, 0, 0, 0};
-    ov::Coordinate new_shape_end{batch_size, num_kv_heads, new_seq_len, head_size};
+    ov::Coordinate new_shape_end{shape};
+
     auto new_tensor = ov::Tensor(tensor, new_shape_begin, new_shape_end);
 
     return new_tensor;