0.00.066.092 I log_info: verbosity = 3 (adjust with the `-lv N` CLI arg)
0.00.066.095 I device_info:
0.00.066.114 I   - ROCm0   : AMD Radeon Graphics (32752 MiB, 32728 MiB free)
0.00.066.117 I   - ROCm1   : AMD Radeon Graphics (32752 MiB, 32728 MiB free)
0.00.066.125 I   - CPU     : AMD Ryzen 9 3900X 12-Core Processor (128730 MiB, 128730 MiB free)
0.00.066.182 I system_info: n_threads = 12 (n_threads_batch = 12) / 24 | ROCm : NO_VMM = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 | 
0.00.066.185 I srv  llama_server: n_parallel is set to auto, using n_parallel = 4 and kv_unified = true
0.00.066.226 I srv          init: running without SSL
0.00.066.294 I srv          init: using 23 threads for HTTP server
0.00.066.629 I srv         start: binding port with default address family
0.00.067.846 I srv  llama_server: loading model
0.00.067.855 I srv    load_model: loading model '/home/qrf/.cache/llama.cpp/granite-4.0-micro-Q4_K_M.gguf'
0.00.067.914 I common_init_result: fitting params to device memory ...
0.00.067.916 I common_init_result: (for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on)
0.01.819.513 W llama_context: n_ctx_seq (8192) < n_ctx_train (131072) -- the full capacity of the model will not be utilized
0.01.851.876 I common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
0.01.985.402 I srv    load_model: initializing slots, n_slots = 4
0.02.033.015 W common_speculative_init: no implementations specified for speculative decoding
0.02.033.020 I slot   load_model: id  0 | task -1 | new slot, n_ctx = 8192
0.02.033.026 I slot   load_model: id  1 | task -1 | new slot, n_ctx = 8192
0.02.033.027 I slot   load_model: id  2 | task -1 | new slot, n_ctx = 8192
0.02.033.027 I slot   load_model: id  3 | task -1 | new slot, n_ctx = 8192
0.02.033.076 I srv    load_model: prompt cache is enabled, size limit: 8192 MiB
0.02.033.076 I srv    load_model: use `--cache-ram 0` to disable the prompt cache
0.02.033.077 I srv    load_model: for more info see https://github.com/ggml-org/llama.cpp/pull/16391
0.02.033.077 I srv    load_model: context checkpoints enabled, max = 32, min spacing = 256
0.02.033.093 I srv          init: idle slots will be saved to prompt cache and cleared upon starting a new task
0.02.040.763 I init: chat template, example_format: '<|start_of_role|>system<|end_of_role|>You are a helpful assistant<|end_of_text|>
<|start_of_role|>user<|end_of_role|>Hello<|end_of_text|>
<|start_of_role|>assistant<|end_of_role|>Hi there<|end_of_text|>
<|start_of_role|>user<|end_of_role|>How are you?<|end_of_text|>
<|start_of_role|>assistant<|end_of_role|>'
0.02.045.564 I srv          init: init: chat template, thinking = 0
0.02.045.600 I srv  llama_server: model loaded
0.02.045.603 I srv  llama_server: server is listening on http://127.0.0.1:8090
0.02.045.608 I srv  update_slots: all slots are idle
0.04.969.301 I srv    operator(): Chat format: peg-native
0.04.969.494 I slot get_availabl: id  3 | task -1 | selected slot by LRU, t_last = -1
0.04.969.497 I srv  get_availabl: updating prompt cache
0.04.969.504 I srv          load:  - looking for better prompt, base f_keep = -1.000, sim = 0.000
0.04.969.508 I srv        update:  - cache state: 0 prompts, 0.000 MiB (limits: 8192.000 MiB, 8192 tokens, 8589934592 est)
0.04.969.509 I srv  get_availabl: prompt cache update took 0.01 ms
0.04.969.580 I slot launch_slot_: id  3 | task 0 | processing task, is_child = 0
0.04.969.582 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.04.969.583 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.04.969.591 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.04.969.592 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.04.969.597 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.04.969.597 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.05.804.819 I slot print_timing: id  3 | task 0 | prompt eval time =     163.78 ms /    72 tokens (    2.27 ms per token,   439.62 tokens per second)
0.05.804.824 I slot print_timing: id  3 | task 0 |        eval time =     671.42 ms /    52 tokens (   12.91 ms per token,    77.45 tokens per second)
0.05.804.824 I slot print_timing: id  3 | task 0 |       total time =     835.20 ms /   124 tokens
0.05.804.827 I slot print_timing: id  3 | task 0 |    graphs reused =         51
0.05.804.882 I slot      release: id  3 | task 0 | stop processing: n_tokens = 123, truncated = 0
0.05.804.891 I srv  update_slots: all slots are idle
0.05.811.080 I srv    operator(): Chat format: peg-native
0.05.811.221 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.766 (> 0.100 thold), f_keep = 0.398
0.05.811.223 I srv  get_availabl: updating prompt cache
0.05.811.246 W srv   prompt_save:  - saving prompt with length 123, total state size = 9.612 MiB (draft: 0.000 MiB)
0.05.820.934 I srv          load:  - looking for better prompt, base f_keep = 0.398, sim = 0.766
0.05.820.939 I srv        update:  - cache state: 1 prompts, 9.612 MiB (limits: 8192.000 MiB, 8192 tokens, 104832 est)
0.05.820.939 I srv        update:    - prompt 0x5f2084b9b770:     123 tokens, checkpoints:  0,     9.612 MiB
0.05.820.940 I srv  get_availabl: prompt cache update took 9.72 ms
0.05.820.999 I slot launch_slot_: id  3 | task 54 | processing task, is_child = 0
0.05.821.001 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.05.821.002 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.05.821.008 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.05.821.008 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.05.821.013 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.05.821.013 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.06.036.786 I slot print_timing: id  3 | task 54 | prompt eval time =      41.81 ms /    15 tokens (    2.79 ms per token,   358.77 tokens per second)
0.06.036.789 I slot print_timing: id  3 | task 54 |        eval time =     173.95 ms /    14 tokens (   12.42 ms per token,    80.48 tokens per second)
0.06.036.790 I slot print_timing: id  3 | task 54 |       total time =     215.76 ms /    29 tokens
0.06.036.790 I slot print_timing: id  3 | task 54 |    graphs reused =         63
0.06.036.829 I slot      release: id  3 | task 54 | stop processing: n_tokens = 77, truncated = 0
0.06.036.834 I srv  update_slots: all slots are idle
0.06.042.934 I srv    operator(): Chat format: peg-native
0.06.043.086 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.742 (> 0.100 thold), f_keep = 0.636
0.06.043.135 I slot launch_slot_: id  3 | task 69 | processing task, is_child = 0
0.06.043.137 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.06.043.137 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.06.043.144 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.06.043.144 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.06.043.156 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.06.043.156 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.06.337.685 I slot print_timing: id  3 | task 69 | prompt eval time =      69.39 ms /    17 tokens (    4.08 ms per token,   244.98 tokens per second)
0.06.337.688 I slot print_timing: id  3 | task 69 |        eval time =     225.12 ms /    18 tokens (   12.51 ms per token,    79.96 tokens per second)
0.06.337.689 I slot print_timing: id  3 | task 69 |       total time =     294.51 ms /    35 tokens
0.06.337.690 I slot print_timing: id  3 | task 69 |    graphs reused =         79
0.06.337.713 I slot      release: id  3 | task 69 | stop processing: n_tokens = 83, truncated = 0
0.06.337.718 I srv  update_slots: all slots are idle
0.06.343.974 I srv    operator(): Chat format: peg-native
0.06.344.114 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.769 (> 0.100 thold), f_keep = 0.602
0.06.344.162 I slot launch_slot_: id  3 | task 88 | processing task, is_child = 0
0.06.344.164 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.06.344.165 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.06.344.171 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.06.344.172 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.06.344.183 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.06.344.184 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.07.117.740 I slot print_timing: id  3 | task 88 | prompt eval time =      41.63 ms /    15 tokens (    2.78 ms per token,   360.28 tokens per second)
0.07.117.743 I slot print_timing: id  3 | task 88 |        eval time =     731.91 ms /    57 tokens (   12.84 ms per token,    77.88 tokens per second)
0.07.117.744 I slot print_timing: id  3 | task 88 |       total time =     773.54 ms /    72 tokens
0.07.117.744 I slot print_timing: id  3 | task 88 |    graphs reused =        134
0.07.117.771 I slot      release: id  3 | task 88 | stop processing: n_tokens = 121, truncated = 0
0.07.117.777 I srv  update_slots: all slots are idle
0.07.124.259 I srv    operator(): Chat format: peg-native
0.07.124.448 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.746 (> 0.100 thold), f_keep = 0.413
0.07.124.451 I srv  get_availabl: updating prompt cache
0.07.124.465 W srv   prompt_save:  - saving prompt with length 121, total state size = 9.455 MiB (draft: 0.000 MiB)
0.07.133.747 I srv          load:  - looking for better prompt, base f_keep = 0.413, sim = 0.746
0.07.133.751 I srv          load:  - found better prompt with f_keep = 0.423, sim = 0.776
0.07.139.854 I srv        update:  - cache state: 1 prompts, 9.455 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.07.139.857 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.07.139.859 I srv  get_availabl: prompt cache update took 15.41 ms
0.07.139.927 I slot launch_slot_: id  3 | task 146 | processing task, is_child = 0
0.07.139.930 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.07.139.931 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.07.139.938 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.07.139.938 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.07.139.943 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.07.139.943 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.07.459.799 I slot print_timing: id  3 | task 146 | prompt eval time =      41.78 ms /    15 tokens (    2.79 ms per token,   358.99 tokens per second)
0.07.459.802 I slot print_timing: id  3 | task 146 |        eval time =     278.06 ms /    22 tokens (   12.64 ms per token,    79.12 tokens per second)
0.07.459.803 I slot print_timing: id  3 | task 146 |       total time =     319.84 ms /    37 tokens
0.07.459.804 I slot print_timing: id  3 | task 146 |    graphs reused =        154
0.07.459.844 I slot      release: id  3 | task 146 | stop processing: n_tokens = 88, truncated = 0
0.07.459.849 I srv  update_slots: all slots are idle
0.07.466.012 I srv    operator(): Chat format: peg-native
0.07.466.158 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.731 (> 0.100 thold), f_keep = 0.557
0.07.466.197 I slot launch_slot_: id  3 | task 169 | processing task, is_child = 0
0.07.466.203 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.07.466.204 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.07.466.210 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.07.466.210 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.07.466.215 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.07.466.216 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.07.710.764 I slot print_timing: id  3 | task 169 | prompt eval time =      69.37 ms /    18 tokens (    3.85 ms per token,   259.47 tokens per second)
0.07.710.768 I slot print_timing: id  3 | task 169 |        eval time =     175.16 ms /    14 tokens (   12.51 ms per token,    79.93 tokens per second)
0.07.710.769 I slot print_timing: id  3 | task 169 |       total time =     244.54 ms /    32 tokens
0.07.710.770 I slot print_timing: id  3 | task 169 |    graphs reused =        166
0.07.710.795 I slot      release: id  3 | task 169 | stop processing: n_tokens = 80, truncated = 0
0.07.710.800 I srv  update_slots: all slots are idle
0.07.716.929 I srv    operator(): Chat format: peg-native
0.07.717.070 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.690 (> 0.100 thold), f_keep = 0.613
0.07.717.101 I slot launch_slot_: id  3 | task 184 | processing task, is_child = 0
0.07.717.102 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.07.717.102 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.07.717.108 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.07.717.108 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.07.717.113 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.07.717.113 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.08.195.764 I slot print_timing: id  3 | task 184 | prompt eval time =      69.55 ms /    22 tokens (    3.16 ms per token,   316.32 tokens per second)
0.08.195.767 I slot print_timing: id  3 | task 184 |        eval time =     409.09 ms /    32 tokens (   12.78 ms per token,    78.22 tokens per second)
0.08.195.768 I slot print_timing: id  3 | task 184 |       total time =     478.64 ms /    54 tokens
0.08.195.768 I slot print_timing: id  3 | task 184 |    graphs reused =        196
0.08.195.795 I slot      release: id  3 | task 184 | stop processing: n_tokens = 102, truncated = 0
0.08.195.801 I srv  update_slots: all slots are idle
0.08.202.076 I srv    operator(): Chat format: peg-native
0.08.202.216 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.820 (> 0.100 thold), f_keep = 0.490
0.08.202.218 I srv  get_availabl: updating prompt cache
0.08.202.237 W srv   prompt_save:  - saving prompt with length 102, total state size = 7.971 MiB (draft: 0.000 MiB)
0.08.210.463 I srv          load:  - looking for better prompt, base f_keep = 0.490, sim = 0.820
0.08.210.468 I srv        update:  - cache state: 2 prompts, 17.426 MiB (limits: 8192.000 MiB, 8192 tokens, 104830 est)
0.08.210.469 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.08.210.470 I srv        update:    - prompt 0x5f2084b9b770:     102 tokens, checkpoints:  0,     7.971 MiB
0.08.210.470 I srv  get_availabl: prompt cache update took 8.25 ms
0.08.210.536 I slot launch_slot_: id  3 | task 217 | processing task, is_child = 0
0.08.210.538 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.08.210.538 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.08.210.544 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.08.210.544 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.08.210.549 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.08.210.549 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.09.023.769 I slot print_timing: id  3 | task 217 | prompt eval time =      41.34 ms /    11 tokens (    3.76 ms per token,   266.09 tokens per second)
0.09.023.773 I slot print_timing: id  3 | task 217 |        eval time =     771.87 ms /    60 tokens (   12.86 ms per token,    77.73 tokens per second)
0.09.023.773 I slot print_timing: id  3 | task 217 |       total time =     813.21 ms /    71 tokens
0.09.023.774 I slot print_timing: id  3 | task 217 |    graphs reused =        254
0.09.023.800 I slot      release: id  3 | task 217 | stop processing: n_tokens = 120, truncated = 0
0.09.023.805 I srv  update_slots: all slots are idle
0.09.030.482 I srv    operator(): Chat format: peg-native
0.09.030.625 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.671 (> 0.100 thold), f_keep = 0.408
0.09.030.628 I srv  get_availabl: updating prompt cache
0.09.030.645 W srv   prompt_save:  - saving prompt with length 120, total state size = 9.377 MiB (draft: 0.000 MiB)
0.09.040.301 I srv          load:  - looking for better prompt, base f_keep = 0.408, sim = 0.671
0.09.040.306 I srv        update:  - cache state: 3 prompts, 26.804 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.09.040.307 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.09.040.308 I srv        update:    - prompt 0x5f2084b9b770:     102 tokens, checkpoints:  0,     7.971 MiB
0.09.040.308 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.09.040.309 I srv  get_availabl: prompt cache update took 9.68 ms
0.09.040.372 I slot launch_slot_: id  3 | task 278 | processing task, is_child = 0
0.09.040.373 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.09.040.373 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.09.040.380 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.09.040.380 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.09.040.385 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.09.040.385 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.09.478.775 I slot print_timing: id  3 | task 278 | prompt eval time =      69.70 ms /    24 tokens (    2.90 ms per token,   344.31 tokens per second)
0.09.478.778 I slot print_timing: id  3 | task 278 |        eval time =     368.67 ms /    29 tokens (   12.71 ms per token,    78.66 tokens per second)
0.09.478.779 I slot print_timing: id  3 | task 278 |       total time =     438.38 ms /    53 tokens
0.09.478.780 I slot print_timing: id  3 | task 278 |    graphs reused =        281
0.09.478.806 I slot      release: id  3 | task 278 | stop processing: n_tokens = 101, truncated = 0
0.09.478.812 I srv  update_slots: all slots are idle
0.09.485.320 I srv    operator(): Chat format: peg-native
0.09.485.470 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.742 (> 0.100 thold), f_keep = 0.485
0.09.485.472 I srv  get_availabl: updating prompt cache
0.09.485.486 W srv   prompt_save:  - saving prompt with length 101, total state size = 7.893 MiB (draft: 0.000 MiB)
0.09.494.174 I srv          load:  - looking for better prompt, base f_keep = 0.485, sim = 0.742
0.09.494.178 I srv          load:  - found better prompt with f_keep = 0.490, sim = 0.758
0.09.498.722 I srv        update:  - cache state: 3 prompts, 26.725 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.09.498.725 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.09.498.726 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.09.498.726 I srv        update:    - prompt 0x5f2085308460:     101 tokens, checkpoints:  0,     7.893 MiB
0.09.498.728 I srv  get_availabl: prompt cache update took 13.25 ms
0.09.498.787 I slot launch_slot_: id  3 | task 308 | processing task, is_child = 0
0.09.498.792 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.09.498.792 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.09.498.798 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.09.498.798 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.09.498.803 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.09.498.803 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.10.028.778 I slot print_timing: id  3 | task 308 | prompt eval time =      43.46 ms /    16 tokens (    2.72 ms per token,   368.19 tokens per second)
0.10.028.781 I slot print_timing: id  3 | task 308 |        eval time =     486.50 ms /    38 tokens (   12.80 ms per token,    78.11 tokens per second)
0.10.028.782 I slot print_timing: id  3 | task 308 |       total time =     529.96 ms /    54 tokens
0.10.028.782 I slot print_timing: id  3 | task 308 |    graphs reused =        317
0.10.028.810 I slot      release: id  3 | task 308 | stop processing: n_tokens = 103, truncated = 0
0.10.028.816 I srv  update_slots: all slots are idle
0.10.035.291 I srv    operator(): Chat format: peg-native
0.10.035.442 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.681 (> 0.100 thold), f_keep = 0.476
0.10.035.444 I srv  get_availabl: updating prompt cache
0.10.035.458 W srv   prompt_save:  - saving prompt with length 103, total state size = 8.049 MiB (draft: 0.000 MiB)
0.10.043.726 I srv          load:  - looking for better prompt, base f_keep = 0.476, sim = 0.681
0.10.043.731 I srv        update:  - cache state: 4 prompts, 34.774 MiB (limits: 8192.000 MiB, 8192 tokens, 104830 est)
0.10.043.731 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.10.043.732 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.10.043.732 I srv        update:    - prompt 0x5f2085308460:     101 tokens, checkpoints:  0,     7.893 MiB
0.10.043.733 I srv        update:    - prompt 0x5f2084accfd0:     103 tokens, checkpoints:  0,     8.049 MiB
0.10.043.733 I srv  get_availabl: prompt cache update took 8.29 ms
0.10.043.781 I slot launch_slot_: id  3 | task 347 | processing task, is_child = 0
0.10.043.781 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.10.043.782 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.10.043.787 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.10.043.788 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.10.043.792 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.10.043.792 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.10.481.788 I slot print_timing: id  3 | task 347 | prompt eval time =      69.27 ms /    23 tokens (    3.01 ms per token,   332.06 tokens per second)
0.10.481.791 I slot print_timing: id  3 | task 347 |        eval time =     368.71 ms /    29 tokens (   12.71 ms per token,    78.65 tokens per second)
0.10.481.792 I slot print_timing: id  3 | task 347 |       total time =     437.98 ms /    52 tokens
0.10.481.792 I slot print_timing: id  3 | task 347 |    graphs reused =        344
0.10.481.818 I slot      release: id  3 | task 347 | stop processing: n_tokens = 100, truncated = 0
0.10.481.825 I srv  update_slots: all slots are idle
0.10.488.012 I srv    operator(): Chat format: peg-native
0.10.488.163 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.731 (> 0.100 thold), f_keep = 0.490
0.10.488.165 I srv  get_availabl: updating prompt cache
0.10.488.183 W srv   prompt_save:  - saving prompt with length 100, total state size = 7.815 MiB (draft: 0.000 MiB)
0.10.493.736 I srv          load:  - looking for better prompt, base f_keep = 0.490, sim = 0.731
0.10.493.742 I srv        update:  - cache state: 5 prompts, 42.589 MiB (limits: 8192.000 MiB, 8192 tokens, 104830 est)
0.10.493.743 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.10.493.743 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.10.493.744 I srv        update:    - prompt 0x5f2085308460:     101 tokens, checkpoints:  0,     7.893 MiB
0.10.493.744 I srv        update:    - prompt 0x5f2084accfd0:     103 tokens, checkpoints:  0,     8.049 MiB
0.10.493.745 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.10.493.745 I srv  get_availabl: prompt cache update took 5.58 ms
0.10.493.806 I slot launch_slot_: id  3 | task 377 | processing task, is_child = 0
0.10.493.806 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.10.493.807 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.10.493.813 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.10.493.813 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.10.493.818 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.10.493.818 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.11.425.762 I slot print_timing: id  3 | task 377 | prompt eval time =      69.06 ms /    18 tokens (    3.84 ms per token,   260.64 tokens per second)
0.11.425.765 I slot print_timing: id  3 | task 377 |        eval time =     862.87 ms /    67 tokens (   12.88 ms per token,    77.65 tokens per second)
0.11.425.765 I slot print_timing: id  3 | task 377 |       total time =     931.93 ms /    85 tokens
0.11.425.766 I slot print_timing: id  3 | task 377 |    graphs reused =        409
0.11.425.792 I slot      release: id  3 | task 377 | stop processing: n_tokens = 133, truncated = 0
0.11.425.797 I srv  update_slots: all slots are idle
0.11.432.052 I srv    operator(): Chat format: peg-native
0.11.432.194 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.721 (> 0.100 thold), f_keep = 0.368
0.11.432.196 I srv  get_availabl: updating prompt cache
0.11.432.216 W srv   prompt_save:  - saving prompt with length 133, total state size = 10.393 MiB (draft: 0.000 MiB)
0.11.442.297 I srv          load:  - looking for better prompt, base f_keep = 0.368, sim = 0.721
0.11.442.304 I srv        update:  - cache state: 6 prompts, 52.982 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.11.442.305 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.11.442.306 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.11.442.306 I srv        update:    - prompt 0x5f2085308460:     101 tokens, checkpoints:  0,     7.893 MiB
0.11.442.306 I srv        update:    - prompt 0x5f2084accfd0:     103 tokens, checkpoints:  0,     8.049 MiB
0.11.442.307 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.11.442.308 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.11.442.308 I srv  get_availabl: prompt cache update took 10.11 ms
0.11.442.383 I slot launch_slot_: id  3 | task 445 | processing task, is_child = 0
0.11.442.384 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.11.442.384 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.11.442.390 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.11.442.390 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.11.442.395 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.11.442.395 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.11.684.800 I slot print_timing: id  3 | task 445 | prompt eval time =      69.37 ms /    19 tokens (    3.65 ms per token,   273.89 tokens per second)
0.11.684.803 I slot print_timing: id  3 | task 445 |        eval time =     173.02 ms /    14 tokens (   12.36 ms per token,    80.91 tokens per second)
0.11.684.804 I slot print_timing: id  3 | task 445 |       total time =     242.39 ms /    33 tokens
0.11.684.804 I slot print_timing: id  3 | task 445 |    graphs reused =        421
0.11.684.831 I slot      release: id  3 | task 445 | stop processing: n_tokens = 81, truncated = 0
0.11.684.837 I srv  update_slots: all slots are idle
0.11.690.969 I srv    operator(): Chat format: peg-native
0.11.691.111 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.731 (> 0.100 thold), f_keep = 0.605
0.11.691.142 I slot launch_slot_: id  3 | task 460 | processing task, is_child = 0
0.11.691.143 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.11.691.143 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.11.691.149 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.11.691.149 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.11.691.159 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.11.691.160 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.12.050.789 I slot print_timing: id  3 | task 460 | prompt eval time =      69.25 ms /    18 tokens (    3.85 ms per token,   259.92 tokens per second)
0.12.050.792 I slot print_timing: id  3 | task 460 |        eval time =     290.36 ms /    23 tokens (   12.62 ms per token,    79.21 tokens per second)
0.12.050.793 I slot print_timing: id  3 | task 460 |       total time =     359.61 ms /    41 tokens
0.12.050.794 I slot print_timing: id  3 | task 460 |    graphs reused =        442
0.12.050.820 I slot      release: id  3 | task 460 | stop processing: n_tokens = 89, truncated = 0
0.12.050.827 I srv  update_slots: all slots are idle
0.12.057.056 I srv    operator(): Chat format: peg-native
0.12.057.200 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.710 (> 0.100 thold), f_keep = 0.551
0.12.057.230 I slot launch_slot_: id  3 | task 484 | processing task, is_child = 0
0.12.057.231 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.12.057.231 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.12.057.237 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.12.057.237 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.12.057.242 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.12.057.242 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.12.390.807 I slot print_timing: id  3 | task 484 | prompt eval time =      69.30 ms /    20 tokens (    3.46 ms per token,   288.61 tokens per second)
0.12.390.810 I slot print_timing: id  3 | task 484 |        eval time =     264.26 ms /    21 tokens (   12.58 ms per token,    79.47 tokens per second)
0.12.390.811 I slot print_timing: id  3 | task 484 |       total time =     333.55 ms /    41 tokens
0.12.390.811 I slot print_timing: id  3 | task 484 |    graphs reused =        461
0.12.390.836 I slot      release: id  3 | task 484 | stop processing: n_tokens = 89, truncated = 0
0.12.390.842 I srv  update_slots: all slots are idle
0.12.397.017 I srv    operator(): Chat format: peg-native
0.12.397.159 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.742 (> 0.100 thold), f_keep = 0.551
0.12.397.210 I slot launch_slot_: id  3 | task 506 | processing task, is_child = 0
0.12.397.212 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.12.397.213 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.12.397.220 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.12.397.220 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.12.397.226 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.12.397.227 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.12.626.807 I slot print_timing: id  3 | task 506 | prompt eval time =      69.30 ms /    17 tokens (    4.08 ms per token,   245.31 tokens per second)
0.12.626.811 I slot print_timing: id  3 | task 506 |        eval time =     160.27 ms /    13 tokens (   12.33 ms per token,    81.12 tokens per second)
0.12.626.812 I slot print_timing: id  3 | task 506 |       total time =     229.56 ms /    30 tokens
0.12.626.813 I slot print_timing: id  3 | task 506 |    graphs reused =        472
0.12.626.839 I slot      release: id  3 | task 506 | stop processing: n_tokens = 78, truncated = 0
0.12.626.845 I srv  update_slots: all slots are idle
0.12.633.532 I srv    operator(): Chat format: peg-native
0.12.633.678 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.710 (> 0.100 thold), f_keep = 0.628
0.12.633.712 I slot launch_slot_: id  3 | task 520 | processing task, is_child = 0
0.12.633.713 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.12.633.714 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.12.633.719 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.12.633.720 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.12.633.725 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.12.633.725 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.13.161.812 I slot print_timing: id  3 | task 520 | prompt eval time =      69.08 ms /    20 tokens (    3.45 ms per token,   289.53 tokens per second)
0.13.161.815 I slot print_timing: id  3 | task 520 |        eval time =     459.00 ms /    36 tokens (   12.75 ms per token,    78.43 tokens per second)
0.13.161.816 I slot print_timing: id  3 | task 520 |       total time =     528.07 ms /    56 tokens
0.13.161.816 I slot print_timing: id  3 | task 520 |    graphs reused =        506
0.13.161.843 I slot      release: id  3 | task 520 | stop processing: n_tokens = 104, truncated = 0
0.13.161.850 I srv  update_slots: all slots are idle
0.13.167.982 I srv    operator(): Chat format: peg-native
0.13.168.133 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.766 (> 0.100 thold), f_keep = 0.471
0.13.168.135 I srv  get_availabl: updating prompt cache
0.13.168.153 W srv   prompt_save:  - saving prompt with length 104, total state size = 8.127 MiB (draft: 0.000 MiB)
0.13.176.569 I srv          load:  - looking for better prompt, base f_keep = 0.471, sim = 0.766
0.13.176.573 I srv          load:  - found better prompt with f_keep = 0.495, sim = 0.781
0.13.180.909 I srv        update:  - cache state: 6 prompts, 53.217 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.13.180.912 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.13.180.913 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.13.180.913 I srv        update:    - prompt 0x5f2084accfd0:     103 tokens, checkpoints:  0,     8.049 MiB
0.13.180.914 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.13.180.914 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.13.180.915 I srv        update:    - prompt 0x5f2085759990:     104 tokens, checkpoints:  0,     8.127 MiB
0.13.180.916 I srv  get_availabl: prompt cache update took 12.78 ms
0.13.180.981 I slot launch_slot_: id  3 | task 557 | processing task, is_child = 0
0.13.180.982 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.13.180.983 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.13.180.989 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.13.180.989 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.13.180.994 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.13.180.994 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.13.603.808 I slot print_timing: id  3 | task 557 | prompt eval time =      41.71 ms /    14 tokens (    2.98 ms per token,   335.66 tokens per second)
0.13.603.811 I slot print_timing: id  3 | task 557 |        eval time =     381.09 ms /    30 tokens (   12.70 ms per token,    78.72 tokens per second)
0.13.603.812 I slot print_timing: id  3 | task 557 |       total time =     422.80 ms /    44 tokens
0.13.603.813 I slot print_timing: id  3 | task 557 |    graphs reused =        534
0.13.603.838 I slot      release: id  3 | task 557 | stop processing: n_tokens = 93, truncated = 0
0.13.603.844 I srv  update_slots: all slots are idle
0.13.610.027 I srv    operator(): Chat format: peg-native
0.13.610.168 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.778 (> 0.100 thold), f_keep = 0.527
0.13.610.199 I slot launch_slot_: id  3 | task 588 | processing task, is_child = 0
0.13.610.200 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.13.610.200 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.13.610.206 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.13.610.206 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.13.610.211 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.13.610.211 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.13.786.831 I slot print_timing: id  3 | task 588 | prompt eval time =      41.56 ms /    14 tokens (    2.97 ms per token,   336.87 tokens per second)
0.13.786.834 I slot print_timing: id  3 | task 588 |        eval time =     135.05 ms /    11 tokens (   12.28 ms per token,    81.45 tokens per second)
0.13.786.835 I slot print_timing: id  3 | task 588 |       total time =     176.61 ms /    25 tokens
0.13.786.835 I slot print_timing: id  3 | task 588 |    graphs reused =        543
0.13.786.860 I slot      release: id  3 | task 588 | stop processing: n_tokens = 73, truncated = 0
0.13.786.865 I srv  update_slots: all slots are idle
0.13.793.086 I srv    operator(): Chat format: peg-native
0.13.793.239 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.781 (> 0.100 thold), f_keep = 0.685
0.13.793.292 I slot launch_slot_: id  3 | task 600 | processing task, is_child = 0
0.13.793.294 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.13.793.295 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.13.793.302 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.13.793.302 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.13.793.307 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.13.793.308 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.14.021.776 I slot print_timing: id  3 | task 600 | prompt eval time =      41.59 ms /    14 tokens (    2.97 ms per token,   336.61 tokens per second)
0.14.021.779 I slot print_timing: id  3 | task 600 |        eval time =     186.86 ms /    15 tokens (   12.46 ms per token,    80.27 tokens per second)
0.14.021.780 I slot print_timing: id  3 | task 600 |       total time =     228.45 ms /    29 tokens
0.14.021.780 I slot print_timing: id  3 | task 600 |    graphs reused =        556
0.14.021.806 I slot      release: id  3 | task 600 | stop processing: n_tokens = 78, truncated = 0
0.14.021.812 I srv  update_slots: all slots are idle
0.14.028.368 I srv    operator(): Chat format: peg-native
0.14.028.511 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.778 (> 0.100 thold), f_keep = 0.628
0.14.028.543 I slot launch_slot_: id  3 | task 616 | processing task, is_child = 0
0.14.028.544 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.14.028.544 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.14.028.550 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.14.028.550 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.14.028.554 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.14.028.555 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.14.672.788 I slot print_timing: id  3 | task 616 | prompt eval time =      41.46 ms /    14 tokens (    2.96 ms per token,   337.67 tokens per second)
0.14.672.792 I slot print_timing: id  3 | task 616 |        eval time =     602.76 ms /    47 tokens (   12.82 ms per token,    77.97 tokens per second)
0.14.672.792 I slot print_timing: id  3 | task 616 |       total time =     644.22 ms /    61 tokens
0.14.672.793 I slot print_timing: id  3 | task 616 |    graphs reused =        601
0.14.672.819 I slot      release: id  3 | task 616 | stop processing: n_tokens = 109, truncated = 0
0.14.672.826 I srv  update_slots: all slots are idle
0.14.679.147 I srv    operator(): Chat format: peg-native
0.14.679.290 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.742 (> 0.100 thold), f_keep = 0.450
0.14.679.292 I srv  get_availabl: updating prompt cache
0.14.679.311 W srv   prompt_save:  - saving prompt with length 109, total state size = 8.518 MiB (draft: 0.000 MiB)
0.14.688.008 I srv          load:  - looking for better prompt, base f_keep = 0.450, sim = 0.742
0.14.688.015 I srv        update:  - cache state: 7 prompts, 61.734 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.14.688.016 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.14.688.016 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.14.688.017 I srv        update:    - prompt 0x5f2084accfd0:     103 tokens, checkpoints:  0,     8.049 MiB
0.14.688.017 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.14.688.018 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.14.688.018 I srv        update:    - prompt 0x5f2085759990:     104 tokens, checkpoints:  0,     8.127 MiB
0.14.688.019 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.14.688.020 I srv  get_availabl: prompt cache update took 8.73 ms
0.14.688.068 I slot launch_slot_: id  3 | task 664 | processing task, is_child = 0
0.14.688.069 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.14.688.069 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.14.688.075 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.14.688.075 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.14.688.080 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.14.688.080 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.14.930.804 I slot print_timing: id  3 | task 664 | prompt eval time =      69.36 ms /    17 tokens (    4.08 ms per token,   245.11 tokens per second)
0.14.930.808 I slot print_timing: id  3 | task 664 |        eval time =     173.35 ms /    14 tokens (   12.38 ms per token,    80.76 tokens per second)
0.14.930.808 I slot print_timing: id  3 | task 664 |       total time =     242.71 ms /    31 tokens
0.14.930.809 I slot print_timing: id  3 | task 664 |    graphs reused =        613
0.14.930.836 I slot      release: id  3 | task 664 | stop processing: n_tokens = 79, truncated = 0
0.14.930.844 I srv  update_slots: all slots are idle
0.14.937.304 I srv    operator(): Chat format: peg-native
0.14.937.479 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.754 (> 0.100 thold), f_keep = 0.620
0.14.937.527 I slot launch_slot_: id  3 | task 679 | processing task, is_child = 0
0.14.937.529 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.14.937.530 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.14.937.538 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.14.937.538 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.14.937.544 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.14.937.545 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.15.362.787 I slot print_timing: id  3 | task 679 | prompt eval time =      43.32 ms /    16 tokens (    2.71 ms per token,   369.35 tokens per second)
0.15.362.791 I slot print_timing: id  3 | task 679 |        eval time =     381.91 ms /    30 tokens (   12.73 ms per token,    78.55 tokens per second)
0.15.362.791 I slot print_timing: id  3 | task 679 |       total time =     425.23 ms /    46 tokens
0.15.362.792 I slot print_timing: id  3 | task 679 |    graphs reused =        641
0.15.362.818 I slot      release: id  3 | task 679 | stop processing: n_tokens = 94, truncated = 0
0.15.362.824 I srv  update_slots: all slots are idle
0.15.369.200 I srv    operator(): Chat format: peg-native
0.15.369.348 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.690 (> 0.100 thold), f_keep = 0.521
0.15.369.401 I slot launch_slot_: id  3 | task 710 | processing task, is_child = 0
0.15.369.403 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.15.369.404 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.15.369.410 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.15.369.410 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.15.369.422 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.15.369.423 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.15.768.800 I slot print_timing: id  3 | task 710 | prompt eval time =      69.38 ms /    22 tokens (    3.15 ms per token,   317.09 tokens per second)
0.15.768.803 I slot print_timing: id  3 | task 710 |        eval time =     329.98 ms /    26 tokens (   12.69 ms per token,    78.79 tokens per second)
0.15.768.804 I slot print_timing: id  3 | task 710 |       total time =     399.36 ms /    48 tokens
0.15.768.804 I slot print_timing: id  3 | task 710 |    graphs reused =        665
0.15.768.831 I slot      release: id  3 | task 710 | stop processing: n_tokens = 96, truncated = 0
0.15.768.837 I srv  update_slots: all slots are idle
0.15.775.302 I srv    operator(): Chat format: peg-native
0.15.775.454 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.700 (> 0.100 thold), f_keep = 0.510
0.15.775.485 I slot launch_slot_: id  3 | task 737 | processing task, is_child = 0
0.15.775.487 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.15.775.488 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.15.775.494 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.15.775.494 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.15.775.499 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.15.775.499 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.16.109.836 I slot print_timing: id  3 | task 737 | prompt eval time =      69.36 ms /    21 tokens (    3.30 ms per token,   302.77 tokens per second)
0.16.109.841 I slot print_timing: id  3 | task 737 |        eval time =     264.97 ms /    21 tokens (   12.62 ms per token,    79.26 tokens per second)
0.16.109.842 I slot print_timing: id  3 | task 737 |       total time =     334.33 ms /    42 tokens
0.16.109.843 I slot print_timing: id  3 | task 737 |    graphs reused =        684
0.16.109.872 I slot      release: id  3 | task 737 | stop processing: n_tokens = 90, truncated = 0
0.16.109.880 I srv  update_slots: all slots are idle
0.16.115.907 I srv    operator(): Chat format: peg-native
0.16.116.046 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.790 (> 0.100 thold), f_keep = 0.544
0.16.116.096 I slot launch_slot_: id  3 | task 759 | processing task, is_child = 0
0.16.116.098 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.16.116.098 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.16.116.104 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.16.116.104 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.16.116.109 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.16.116.109 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.16.304.843 I slot print_timing: id  3 | task 759 | prompt eval time =      41.62 ms /    13 tokens (    3.20 ms per token,   312.31 tokens per second)
0.16.304.847 I slot print_timing: id  3 | task 759 |        eval time =     147.10 ms /    12 tokens (   12.26 ms per token,    81.58 tokens per second)
0.16.304.848 I slot print_timing: id  3 | task 759 |       total time =     188.72 ms /    25 tokens
0.16.304.848 I slot print_timing: id  3 | task 759 |    graphs reused =        694
0.16.304.874 I slot      release: id  3 | task 759 | stop processing: n_tokens = 73, truncated = 0
0.16.304.880 I srv  update_slots: all slots are idle
0.16.311.149 I srv    operator(): Chat format: peg-native
0.16.311.295 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.754 (> 0.100 thold), f_keep = 0.671
0.16.311.332 I slot launch_slot_: id  3 | task 772 | processing task, is_child = 0
0.16.311.332 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.16.311.333 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.16.311.338 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.16.311.338 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.16.311.343 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.16.311.343 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.16.553.829 I slot print_timing: id  3 | task 772 | prompt eval time =      43.23 ms /    16 tokens (    2.70 ms per token,   370.11 tokens per second)
0.16.553.832 I slot print_timing: id  3 | task 772 |        eval time =     199.24 ms /    16 tokens (   12.45 ms per token,    80.30 tokens per second)
0.16.553.833 I slot print_timing: id  3 | task 772 |       total time =     242.47 ms /    32 tokens
0.16.553.834 I slot print_timing: id  3 | task 772 |    graphs reused =        708
0.16.553.860 I slot      release: id  3 | task 772 | stop processing: n_tokens = 80, truncated = 0
0.16.553.866 I srv  update_slots: all slots are idle
0.16.559.920 I srv    operator(): Chat format: peg-native
0.16.560.060 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.731 (> 0.100 thold), f_keep = 0.613
0.16.560.109 I slot launch_slot_: id  3 | task 789 | processing task, is_child = 0
0.16.560.111 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.16.560.112 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.16.560.117 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.16.560.117 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.16.560.122 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.16.560.122 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.16.829.841 I slot print_timing: id  3 | task 789 | prompt eval time =      69.33 ms /    18 tokens (    3.85 ms per token,   259.63 tokens per second)
0.16.829.845 I slot print_timing: id  3 | task 789 |        eval time =     200.38 ms /    16 tokens (   12.52 ms per token,    79.85 tokens per second)
0.16.829.846 I slot print_timing: id  3 | task 789 |       total time =     269.71 ms /    34 tokens
0.16.829.847 I slot print_timing: id  3 | task 789 |    graphs reused =        722
0.16.829.874 I slot      release: id  3 | task 789 | stop processing: n_tokens = 82, truncated = 0
0.16.829.880 I srv  update_slots: all slots are idle
0.16.836.221 I srv    operator(): Chat format: peg-native
0.16.836.368 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.636 (> 0.100 thold), f_keep = 0.598
0.16.836.399 I slot launch_slot_: id  3 | task 806 | processing task, is_child = 0
0.16.836.401 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.16.836.401 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.16.836.407 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.16.836.407 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.16.836.418 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.16.836.418 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.18.078.835 I slot print_timing: id  3 | task 806 | prompt eval time =      80.69 ms /    28 tokens (    2.88 ms per token,   346.99 tokens per second)
0.18.078.839 I slot print_timing: id  3 | task 806 |        eval time =    1161.71 ms /    90 tokens (   12.91 ms per token,    77.47 tokens per second)
0.18.078.839 I slot print_timing: id  3 | task 806 |       total time =    1242.40 ms /   118 tokens
0.18.078.840 I slot print_timing: id  3 | task 806 |    graphs reused =        810
0.18.078.867 I slot      release: id  3 | task 806 | stop processing: n_tokens = 166, truncated = 0
0.18.078.874 I srv  update_slots: all slots are idle
0.18.085.218 I srv    operator(): Chat format: peg-native
0.18.085.375 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.725 (> 0.100 thold), f_keep = 0.301
0.18.085.377 I srv  get_availabl: updating prompt cache
0.18.085.395 W srv   prompt_save:  - saving prompt with length 166, total state size = 12.972 MiB (draft: 0.000 MiB)
0.18.097.114 I srv          load:  - looking for better prompt, base f_keep = 0.301, sim = 0.725
0.18.097.121 I srv        update:  - cache state: 8 prompts, 74.706 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.18.097.122 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.18.097.123 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.18.097.123 I srv        update:    - prompt 0x5f2084accfd0:     103 tokens, checkpoints:  0,     8.049 MiB
0.18.097.124 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.18.097.124 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.18.097.125 I srv        update:    - prompt 0x5f2085759990:     104 tokens, checkpoints:  0,     8.127 MiB
0.18.097.125 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.18.097.126 I srv        update:    - prompt 0x5f2082a50ad0:     166 tokens, checkpoints:  0,    12.972 MiB
0.18.097.127 I srv  get_availabl: prompt cache update took 11.75 ms
0.18.097.194 I slot launch_slot_: id  3 | task 897 | processing task, is_child = 0
0.18.097.195 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.18.097.195 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.18.097.202 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.18.097.202 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.18.097.207 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.18.097.207 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.18.600.843 I slot print_timing: id  3 | task 897 | prompt eval time =      69.55 ms /    19 tokens (    3.66 ms per token,   273.18 tokens per second)
0.18.600.846 I slot print_timing: id  3 | task 897 |        eval time =     434.07 ms /    34 tokens (   12.77 ms per token,    78.33 tokens per second)
0.18.600.846 I slot print_timing: id  3 | task 897 |       total time =     503.62 ms /    53 tokens
0.18.600.847 I slot print_timing: id  3 | task 897 |    graphs reused =        842
0.18.600.875 I slot      release: id  3 | task 897 | stop processing: n_tokens = 102, truncated = 0
0.18.600.882 I srv  update_slots: all slots are idle
0.18.607.173 I srv    operator(): Chat format: peg-native
0.18.607.316 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.766 (> 0.100 thold), f_keep = 0.480
0.18.607.318 I srv  get_availabl: updating prompt cache
0.18.607.347 W srv   prompt_save:  - saving prompt with length 102, total state size = 7.971 MiB (draft: 0.000 MiB)
0.18.615.598 I srv          load:  - looking for better prompt, base f_keep = 0.480, sim = 0.766
0.18.615.604 I srv        update:  - cache state: 9 prompts, 82.677 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.18.615.605 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.18.615.605 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.18.615.606 I srv        update:    - prompt 0x5f2084accfd0:     103 tokens, checkpoints:  0,     8.049 MiB
0.18.615.606 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.18.615.607 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.18.615.607 I srv        update:    - prompt 0x5f2085759990:     104 tokens, checkpoints:  0,     8.127 MiB
0.18.615.608 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.18.615.608 I srv        update:    - prompt 0x5f2082a50ad0:     166 tokens, checkpoints:  0,    12.972 MiB
0.18.615.609 I srv        update:    - prompt 0x5f2084cce210:     102 tokens, checkpoints:  0,     7.971 MiB
0.18.615.609 I srv  get_availabl: prompt cache update took 8.29 ms
0.18.615.661 I slot launch_slot_: id  3 | task 932 | processing task, is_child = 0
0.18.615.662 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.18.615.662 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.18.615.668 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.18.615.668 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.18.615.672 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.18.615.673 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.19.312.335 I slot print_timing: id  3 | task 932 | prompt eval time =      41.49 ms /    15 tokens (    2.77 ms per token,   361.57 tokens per second)
0.19.312.340 I slot print_timing: id  3 | task 932 |        eval time =     655.15 ms /    51 tokens (   12.85 ms per token,    77.85 tokens per second)
0.19.312.341 I slot print_timing: id  3 | task 932 |       total time =     696.63 ms /    66 tokens
0.19.312.341 I slot print_timing: id  3 | task 932 |    graphs reused =        891
0.19.312.401 I slot      release: id  3 | task 932 | stop processing: n_tokens = 114, truncated = 0
0.19.312.412 I srv  update_slots: all slots are idle
0.19.318.515 I srv    operator(): Chat format: peg-native
0.19.318.676 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.710 (> 0.100 thold), f_keep = 0.430
0.19.318.679 I srv  get_availabl: updating prompt cache
0.19.318.705 W srv   prompt_save:  - saving prompt with length 114, total state size = 8.908 MiB (draft: 0.000 MiB)
0.19.328.314 I srv          load:  - looking for better prompt, base f_keep = 0.430, sim = 0.710
0.19.328.318 I srv          load:  - found better prompt with f_keep = 0.485, sim = 0.725
0.19.332.982 I srv        update:  - cache state: 9 prompts, 83.536 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.19.332.986 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.19.332.987 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.19.332.987 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.19.332.988 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.19.332.988 I srv        update:    - prompt 0x5f2085759990:     104 tokens, checkpoints:  0,     8.127 MiB
0.19.332.989 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.19.332.989 I srv        update:    - prompt 0x5f2082a50ad0:     166 tokens, checkpoints:  0,    12.972 MiB
0.19.332.990 I srv        update:    - prompt 0x5f2084cce210:     102 tokens, checkpoints:  0,     7.971 MiB
0.19.332.990 I srv        update:    - prompt 0x5f2084b6fec0:     114 tokens, checkpoints:  0,     8.908 MiB
0.19.332.991 I srv  get_availabl: prompt cache update took 14.31 ms
0.19.333.058 I slot launch_slot_: id  3 | task 984 | processing task, is_child = 0
0.19.333.059 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.19.333.059 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.19.333.065 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.19.333.066 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.19.333.070 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.19.333.070 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.19.837.886 I slot print_timing: id  3 | task 984 | prompt eval time =      69.69 ms /    19 tokens (    3.67 ms per token,   272.62 tokens per second)
0.19.837.889 I slot print_timing: id  3 | task 984 |        eval time =     435.11 ms /    34 tokens (   12.80 ms per token,    78.14 tokens per second)
0.19.837.890 I slot print_timing: id  3 | task 984 |       total time =     504.80 ms /    53 tokens
0.19.837.890 I slot print_timing: id  3 | task 984 |    graphs reused =        923
0.19.837.919 I slot      release: id  3 | task 984 | stop processing: n_tokens = 102, truncated = 0
0.19.837.925 I srv  update_slots: all slots are idle
0.19.844.019 I srv    operator(): Chat format: peg-native
0.19.844.162 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.694 (> 0.100 thold), f_keep = 0.490
0.19.844.165 I srv  get_availabl: updating prompt cache
0.19.844.183 W srv   prompt_save:  - saving prompt with length 102, total state size = 7.971 MiB (draft: 0.000 MiB)
0.19.849.670 I srv          load:  - looking for better prompt, base f_keep = 0.490, sim = 0.694
0.19.849.673 I srv          load:  - found better prompt with f_keep = 0.500, sim = 0.722
0.19.854.511 I srv        update:  - cache state: 9 prompts, 83.380 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.19.854.514 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.19.854.514 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.19.854.515 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.19.854.516 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.19.854.516 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.19.854.517 I srv        update:    - prompt 0x5f2082a50ad0:     166 tokens, checkpoints:  0,    12.972 MiB
0.19.854.517 I srv        update:    - prompt 0x5f2084cce210:     102 tokens, checkpoints:  0,     7.971 MiB
0.19.854.518 I srv        update:    - prompt 0x5f2084b6fec0:     114 tokens, checkpoints:  0,     8.908 MiB
0.19.854.518 I srv        update:    - prompt 0x5f2084ad2730:     102 tokens, checkpoints:  0,     7.971 MiB
0.19.854.519 I srv  get_availabl: prompt cache update took 10.35 ms
0.19.854.584 I slot launch_slot_: id  3 | task 1019 | processing task, is_child = 0
0.19.854.585 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.19.854.585 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.19.854.590 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.19.854.591 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.19.854.595 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.19.854.595 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.20.162.899 I slot print_timing: id  3 | task 1019 | prompt eval time =      69.26 ms /    20 tokens (    3.46 ms per token,   288.78 tokens per second)
0.20.162.903 I slot print_timing: id  3 | task 1019 |        eval time =     239.03 ms /    19 tokens (   12.58 ms per token,    79.49 tokens per second)
0.20.162.903 I slot print_timing: id  3 | task 1019 |       total time =     308.29 ms /    39 tokens
0.20.162.904 I slot print_timing: id  3 | task 1019 |    graphs reused =        940
0.20.162.931 I slot      release: id  3 | task 1019 | stop processing: n_tokens = 90, truncated = 0
0.20.162.938 I srv  update_slots: all slots are idle
0.20.169.017 I srv    operator(): Chat format: peg-native
0.20.169.159 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.766 (> 0.100 thold), f_keep = 0.544
0.20.169.193 I slot launch_slot_: id  3 | task 1039 | processing task, is_child = 0
0.20.169.193 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.20.169.194 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.20.169.199 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.20.169.200 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.20.169.204 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.20.169.205 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.20.761.840 I slot print_timing: id  3 | task 1039 | prompt eval time =      41.70 ms /    15 tokens (    2.78 ms per token,   359.69 tokens per second)
0.20.761.843 I slot print_timing: id  3 | task 1039 |        eval time =     550.92 ms /    43 tokens (   12.81 ms per token,    78.05 tokens per second)
0.20.761.844 I slot print_timing: id  3 | task 1039 |       total time =     592.62 ms /    58 tokens
0.20.761.845 I slot print_timing: id  3 | task 1039 |    graphs reused =        981
0.20.761.872 I slot      release: id  3 | task 1039 | stop processing: n_tokens = 106, truncated = 0
0.20.761.878 I srv  update_slots: all slots are idle
0.20.767.996 I srv    operator(): Chat format: peg-native
0.20.768.140 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.817 (> 0.100 thold), f_keep = 0.462
0.20.768.142 I srv  get_availabl: updating prompt cache
0.20.768.157 W srv   prompt_save:  - saving prompt with length 106, total state size = 8.283 MiB (draft: 0.000 MiB)
0.20.773.607 I srv          load:  - looking for better prompt, base f_keep = 0.462, sim = 0.817
0.20.773.612 I srv        update:  - cache state: 10 prompts, 91.663 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.20.773.613 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.20.773.614 I srv        update:    - prompt 0x5f2082ae6830:     120 tokens, checkpoints:  0,     9.377 MiB
0.20.773.614 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.20.773.615 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.20.773.615 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.20.773.616 I srv        update:    - prompt 0x5f2082a50ad0:     166 tokens, checkpoints:  0,    12.972 MiB
0.20.773.616 I srv        update:    - prompt 0x5f2084cce210:     102 tokens, checkpoints:  0,     7.971 MiB
0.20.773.617 I srv        update:    - prompt 0x5f2084b6fec0:     114 tokens, checkpoints:  0,     8.908 MiB
0.20.773.617 I srv        update:    - prompt 0x5f2084ad2730:     102 tokens, checkpoints:  0,     7.971 MiB
0.20.773.618 I srv        update:    - prompt 0x5f2084accfd0:     106 tokens, checkpoints:  0,     8.283 MiB
0.20.773.618 I srv  get_availabl: prompt cache update took 5.48 ms
0.20.773.660 I slot launch_slot_: id  3 | task 1083 | processing task, is_child = 0
0.20.773.661 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.20.773.662 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.20.773.669 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.20.773.669 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.20.773.675 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.20.773.676 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.21.040.860 I slot print_timing: id  3 | task 1083 | prompt eval time =      41.27 ms /    11 tokens (    3.75 ms per token,   266.54 tokens per second)
0.21.040.863 I slot print_timing: id  3 | task 1083 |        eval time =     225.90 ms /    18 tokens (   12.55 ms per token,    79.68 tokens per second)
0.21.040.863 I slot print_timing: id  3 | task 1083 |       total time =     267.17 ms /    29 tokens
0.21.040.864 I slot print_timing: id  3 | task 1083 |    graphs reused =        997
0.21.040.891 I slot      release: id  3 | task 1083 | stop processing: n_tokens = 77, truncated = 0
0.21.040.897 I srv  update_slots: all slots are idle
0.21.047.044 I srv    operator(): Chat format: peg-native
0.21.047.183 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.721 (> 0.100 thold), f_keep = 0.636
0.21.047.230 I slot launch_slot_: id  3 | task 1102 | processing task, is_child = 0
0.21.047.232 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.21.047.232 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.21.047.238 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.21.047.238 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.21.047.249 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.21.047.250 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.21.848.872 I slot print_timing: id  3 | task 1102 | prompt eval time =      69.40 ms /    19 tokens (    3.65 ms per token,   273.76 tokens per second)
0.21.848.875 I slot print_timing: id  3 | task 1102 |        eval time =     732.20 ms /    57 tokens (   12.85 ms per token,    77.85 tokens per second)
0.21.848.876 I slot print_timing: id  3 | task 1102 |       total time =     801.61 ms /    76 tokens
0.21.848.877 I slot print_timing: id  3 | task 1102 |    graphs reused =       1052
0.21.848.902 I slot      release: id  3 | task 1102 | stop processing: n_tokens = 124, truncated = 0
0.21.848.908 I srv  update_slots: all slots are idle
0.21.855.132 I srv    operator(): Chat format: peg-native
0.21.855.275 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.766 (> 0.100 thold), f_keep = 0.395
0.21.855.278 I srv  get_availabl: updating prompt cache
0.21.855.295 W srv   prompt_save:  - saving prompt with length 124, total state size = 9.690 MiB (draft: 0.000 MiB)
0.21.864.694 I srv          load:  - looking for better prompt, base f_keep = 0.395, sim = 0.766
0.21.864.700 I srv          load:  - found better prompt with f_keep = 0.425, sim = 0.797
0.21.869.581 I srv        update:  - cache state: 10 prompts, 91.976 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.21.869.583 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.21.869.584 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.21.869.585 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.21.869.585 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.21.869.586 I srv        update:    - prompt 0x5f2082a50ad0:     166 tokens, checkpoints:  0,    12.972 MiB
0.21.869.586 I srv        update:    - prompt 0x5f2084cce210:     102 tokens, checkpoints:  0,     7.971 MiB
0.21.869.587 I srv        update:    - prompt 0x5f2084b6fec0:     114 tokens, checkpoints:  0,     8.908 MiB
0.21.869.587 I srv        update:    - prompt 0x5f2084ad2730:     102 tokens, checkpoints:  0,     7.971 MiB
0.21.869.588 I srv        update:    - prompt 0x5f2084accfd0:     106 tokens, checkpoints:  0,     8.283 MiB
0.21.869.588 I srv        update:    - prompt 0x5f2084a52330:     124 tokens, checkpoints:  0,     9.690 MiB
0.21.869.590 I srv  get_availabl: prompt cache update took 14.31 ms
0.21.869.650 I slot launch_slot_: id  3 | task 1160 | processing task, is_child = 0
0.21.869.651 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.21.869.651 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.21.869.657 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.21.869.657 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.21.869.662 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.21.869.662 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.22.981.893 I slot print_timing: id  3 | task 1160 | prompt eval time =      41.37 ms /    13 tokens (    3.18 ms per token,   314.21 tokens per second)
0.22.981.897 I slot print_timing: id  3 | task 1160 |        eval time =    1070.84 ms /    83 tokens (   12.90 ms per token,    77.51 tokens per second)
0.22.981.897 I slot print_timing: id  3 | task 1160 |       total time =    1112.22 ms /    96 tokens
0.22.981.898 I slot print_timing: id  3 | task 1160 |    graphs reused =       1133
0.22.981.924 I slot      release: id  3 | task 1160 | stop processing: n_tokens = 146, truncated = 0
0.22.981.930 I srv  update_slots: all slots are idle
0.22.988.092 I srv    operator(): Chat format: peg-native
0.22.988.231 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.794 (> 0.100 thold), f_keep = 0.342
0.22.988.231 I srv  get_availabl: updating prompt cache
0.22.988.249 W srv   prompt_save:  - saving prompt with length 146, total state size = 11.409 MiB (draft: 0.000 MiB)
0.22.998.936 I srv          load:  - looking for better prompt, base f_keep = 0.342, sim = 0.794
0.22.998.944 I srv        update:  - cache state: 11 prompts, 103.385 MiB (limits: 8192.000 MiB, 8192 tokens, 104831 est)
0.22.998.945 I srv        update:    - prompt 0x5f208518b260:     121 tokens, checkpoints:  0,     9.455 MiB
0.22.998.945 I srv        update:    - prompt 0x5f2084b681f0:     100 tokens, checkpoints:  0,     7.815 MiB
0.22.998.946 I srv        update:    - prompt 0x5f2084b9b770:     133 tokens, checkpoints:  0,    10.393 MiB
0.22.998.946 I srv        update:    - prompt 0x5f2082b013f0:     109 tokens, checkpoints:  0,     8.518 MiB
0.22.998.947 I srv        update:    - prompt 0x5f2082a50ad0:     166 tokens, checkpoints:  0,    12.972 MiB
0.22.998.947 I srv        update:    - prompt 0x5f2084cce210:     102 tokens, checkpoints:  0,     7.971 MiB
0.22.998.948 I srv        update:    - prompt 0x5f2084b6fec0:     114 tokens, checkpoints:  0,     8.908 MiB
0.22.998.948 I srv        update:    - prompt 0x5f2084ad2730:     102 tokens, checkpoints:  0,     7.971 MiB
0.22.998.948 I srv        update:    - prompt 0x5f2084accfd0:     106 tokens, checkpoints:  0,     8.283 MiB
0.22.998.949 I srv        update:    - prompt 0x5f2084a52330:     124 tokens, checkpoints:  0,     9.690 MiB
0.22.998.949 I srv        update:    - prompt 0x5f2084a599b0:     146 tokens, checkpoints:  0,    11.409 MiB
0.22.998.950 I srv  get_availabl: prompt cache update took 10.72 ms
0.22.999.008 I slot launch_slot_: id  3 | task 1244 | processing task, is_child = 0
0.22.999.009 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.22.999.010 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.22.999.016 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.22.999.016 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.22.999.021 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.22.999.021 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.23.344.879 I slot print_timing: id  3 | task 1244 | prompt eval time =      41.69 ms /    13 tokens (    3.21 ms per token,   311.81 tokens per second)
0.23.344.883 I slot print_timing: id  3 | task 1244 |        eval time =     304.15 ms /    24 tokens (   12.67 ms per token,    78.91 tokens per second)
0.23.344.883 I slot print_timing: id  3 | task 1244 |       total time =     345.84 ms /    37 tokens
0.23.344.884 I slot print_timing: id  3 | task 1244 |    graphs reused =       1155
0.23.344.909 I slot      release: id  3 | task 1244 | stop processing: n_tokens = 86, truncated = 0
0.23.344.915 I srv  update_slots: all slots are idle
0.23.350.986 I srv    operator(): Chat format: peg-native
0.23.351.125 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.742 (> 0.100 thold), f_keep = 0.570
0.23.351.175 I slot launch_slot_: id  3 | task 1269 | processing task, is_child = 0
0.23.351.176 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.23.351.177 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.23.351.182 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.23.351.183 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.23.351.187 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.23.351.187 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.23.606.926 I slot print_timing: id  3 | task 1269 | prompt eval time =      69.32 ms /    17 tokens (    4.08 ms per token,   245.25 tokens per second)
0.23.606.929 I slot print_timing: id  3 | task 1269 |        eval time =     186.41 ms /    15 tokens (   12.43 ms per token,    80.47 tokens per second)
0.23.606.930 I slot print_timing: id  3 | task 1269 |       total time =     255.73 ms /    32 tokens
0.23.606.931 I slot print_timing: id  3 | task 1269 |    graphs reused =       1168
0.23.606.955 I slot      release: id  3 | task 1269 | stop processing: n_tokens = 80, truncated = 0
0.23.606.961 I srv  update_slots: all slots are idle
0.23.612.922 I srv    operator(): Chat format: peg-native
0.23.613.064 I slot get_availabl: id  3 | task -1 | selected slot by LCP similarity, sim_best = 0.790 (> 0.100 thold), f_keep = 0.613
0.23.613.102 I slot launch_slot_: id  3 | task 1285 | processing task, is_child = 0
0.23.613.103 I slot process_sing: id  0 | task -1 | saving idle slot to prompt cache
0.23.613.104 I slot prompt_clear: id  0 | task -1 | clearing prompt with 0 tokens
0.23.613.111 I slot process_sing: id  1 | task -1 | saving idle slot to prompt cache
0.23.613.111 I slot prompt_clear: id  1 | task -1 | clearing prompt with 0 tokens
0.23.613.117 I slot process_sing: id  2 | task -1 | saving idle slot to prompt cache
0.23.613.117 I slot prompt_clear: id  2 | task -1 | clearing prompt with 0 tokens
0.24.322.891 I slot print_timing: id  3 | task 1285 | prompt eval time =      41.65 ms /    13 tokens (    3.20 ms per token,   312.12 tokens per second)
0.24.322.895 I slot print_timing: id  3 | task 1285 |        eval time =     668.11 ms /    52 tokens (   12.85 ms per token,    77.83 tokens per second)
0.24.322.895 I slot print_timing: id  3 | task 1285 |       total time =     709.76 ms /    65 tokens
0.24.322.896 I slot print_timing: id  3 | task 1285 |    graphs reused =       1218
0.24.322.923 I slot      release: id  3 | task 1285 | stop processing: n_tokens = 113, truncated = 0
0.24.322.928 I srv  update_slots: all slots are idle
0.24.323.389 I srv    operator(): operator(): cleaning up before exit...