| COMMAND: |
| ["/content/Prism-llama.cpp/build/bin/llama-completion", "-m", "/content/models/Bonsai-27B-Q1/Bonsai-27B-Q1_0.gguf", "-p", "User: Explain in one sentence why deterministic checkpoint resume matters.\nAssistant:", "-n", "16", "--seed", "1234", "--temp", "0", "--ctx-size", "256", "-ngl", "999", "--lora", "/content/prism_native_q1_lora/step10_training_pipeline/artifacts/accepted/step10_adapter_final.gguf", "--threads", "2", "--threads-batch", "2", "--batch-size", "32", "--ubatch-size", "16", "--no-display-prompt", "--no-warmup", "--ignore-eos", "-no-cnv"] |
|
|
| STDOUT: |
|
|
|
|
| <think> |
| Here's a thinking process: |
|
|
| 1. **Analyze |
|
|
|
|
|
|
| STDERR: |
| 0.00.019.837 I llama_completion: llama backend init |
| 0.00.019.852 I llama_completion: load the model and apply lora adapter, if any |
| 0.00.019.863 I common_init_result: fitting params to device memory ... |
| 0.00.019.864 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) |
| 0.03.037.031 W llama_context: n_ctx_seq (256) < n_ctx_train (262144) -- the full capacity of the model will not be utilized |
| 0.03.507.553 I llama_completion: llama threadpool init, n_threads = 2 |
| 0.03.509.158 I |
| 0.03.509.265 I system_info: n_threads = 2 (n_threads_batch = 2) / 12 | CUDA : ARCHS = 890 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | AVX512 = 1 | AVX512_VNNI = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 | |
| 0.03.509.270 I |
| 0.03.509.472 I sampler seed: 1234 |
| 0.03.509.487 I sampler params: |
| repeat_last_n = 64, repeat_penalty = 1.000, frequency_penalty = 0.000, presence_penalty = 0.000 |
| dry_multiplier = 0.000, dry_base = 1.750, dry_allowed_length = 2, dry_penalty_last_n = -1 |
| top_k = 20, top_p = 0.950, min_p = 0.050, xtc_probability = 0.000, xtc_threshold = 0.100, typical_p = 1.000, top_n_sigma = -1.000, temp = 0.000 |
| mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000, adaptive_target = -1.000, adaptive_decay = 0.900 |
| 0.03.509.497 I sampler chain: logits -> logit-bias -> ?penalties -> ?dry -> ?top-n-sigma -> top-k -> ?typical -> top-p -> min-p -> ?xtc -> temp-ext -> dist |
| 0.03.509.500 I generate: n_ctx = 256, n_batch = 32, n_predict = 16, n_keep = 0 |
| 0.03.509.501 I |
| 0.04.444.968 I common_perf_print: sampling time = 13.59 ms |
| 0.04.444.969 I common_perf_print: samplers time = 3.87 ms / 31 tokens |
| 0.04.444.974 I common_perf_print: load time = 2.02 ms |
| 0.04.444.976 I common_perf_print: prompt eval time = 0.02 ms / 15 tokens ( 0.00 ms per token, 681818.18 tokens per second) |
| 0.04.444.979 I common_perf_print: eval time = 449.21 ms / 15 runs ( 29.95 ms per token, 33.39 tokens per second) |
| 0.04.444.980 I common_perf_print: total time = 937.42 ms / 30 tokens |
| 0.04.444.981 I common_perf_print: unaccounted time = 474.60 ms / 50.6 % (total - sampling - prompt eval - eval) / (total) |
| 0.04.444.984 I common_perf_print: graphs reused = 14 |
|
|
|
|
