Dynamic KV cache allocation #6027

Job	Run time
cpp-beam_search_causal_lm-ubuntu (./build/samples/cpp/beam_search_causal_lm/beam_search_causal_lm)	3m 48s
cpp-beam_search_causal_lm-ubuntu (python ./samples/python/beam_search_causal_lm/beam_search_causa...	3m 33s
cpp-beam_search_causal_lm-Qwen1_5-7B-Chat	3m 30s
cpp-beam_search_causal_lm-Phi-2	3m 37s
cpp-greedy_causal_lm-Qwen-7B-Chat	3m 38s
cpp-beam_search_causal_lm-notus-7b-v1	3m 27s
cpp-greedy_causal_lm-windows	3m 24s
cpp-multinomial-greedy_causal_lm-ubuntu	3m 42s
cpp-chat_sample-ubuntu	3m 37s
cpp-continuous-batching-macos	3m 51s
cpp-continuous-batching-ubuntu	3m 39s
cpp-continuous-batching-windows	3m 23s
cpp-greedy_causal_lm-redpajama-3b-chat	3m 34s
cpp-Phi-1_5	3m 19s
cpp-prompt_lookup_decoding_lm-ubuntu	3m 16s
cpp-speculative_decoding_lm-ubuntu	3m 18s
visual_language_chat_sample-ubuntu-internvl2	3m 36s
visual_language_chat_sample-ubuntu-minicpm_v2_6	3m 31s
visual_language_chat_sample-ubuntu-llava_1_5 / visual_language_chat_sample-ubuntu-llava	3m 36s
visual_language_chat_sample-ubuntu-llava_next / visual_language_chat_sample-ubuntu-llava	3m 28s
ci/gha_overall_status_causal_lm	1s
	1h 10m 48s

Provide feedback