\u6d4b\u8bd5\u73af\u5883 windows, llama.cpp, cuda, \u8fd0\u884c\u547d\u4ee4\uff1a<\/p>\n\n\n\n
main.exe -t 16 -ngl 256 -m models\\grok-1-q2_k_s-00001-of-00009.gguf --color -c 4096 --temp 0.9 --repeat_penalty 1.1 -n -1 -p \"### Instruction: Write me a linked list implementation in C\/C++, starting with '```' and ending with '```'\\n### Response:\"\n<\/pre><\/div>\n\n\n\nmain.exe -t 16 -ngl 256 -m models\\grok-1-q2_k_s-00001-of-00009.gguf --color -c 4096 --temp 0.9 --repeat_penalty 1.1 -n -1 -p \"### Instruction: Write me a linked list implementation in C\/C++, starting with '```' and ending with '```'\\n### Response:\"\nLog start\nmain: build = 2555 (d0e2f641)\nmain: built with MSVC 19.39.33523.0 for x64\nmain: seed = 1711616424\nllama_model_loader: additional 8 GGUFs metadata loaded.\nllama_model_loader: loaded meta data with 27 key-value pairs and 2114 tensors from c:\\ai\\grok\\grok-1\\models\\grok-1-q2_k_s-00001-of-00009.gguf (version GGUF V3 (latest))\nllama_model_loader: Dumping metadata keys\/values. Note: KV overrides do not apply in this output.\nllama_model_loader: - kv 0: general.architecture str = grok\nllama_model_loader: - kv 1: general.name str = Grok\nllama_model_loader: - kv 2: grok.block_count u32 = 64\nllama_model_loader: - kv 3: grok.context_length u32 = 4096\nllama_model_loader: - kv 4: grok.embedding_length u32 = 6144\nllama_model_loader: - kv 5: grok.feed_forward_length u32 = 32768\nllama_model_loader: - kv 6: grok.attention.head_count u32 = 48\nllama_model_loader: - kv 7: grok.attention.head_count_kv u32 = 8\nllama_model_loader: - kv 8: grok.rope.freq_base f32 = 100000.000000\nllama_model_loader: - kv 9: grok.attention.layer_norm_rms_epsilon f32 = 0.000010\nllama_model_loader: - kv 10: grok.expert_count u32 = 8\nllama_model_loader: - kv 11: grok.expert_used_count u32 = 2\nllama_model_loader: - kv 12: general.file_type u32 = 10\nllama_model_loader: - kv 13: tokenizer.ggml.model str = llama\nllama_model_loader: - kv 14: tokenizer.ggml.tokens arr[str,131072] = [\"[PAD]\", \"[BOS]\", \"[EOS]\", \"[UNK]\", ...\nllama_model_loader: - kv 15: tokenizer.ggml.scores arr[f32,131072] = [0.000000, 0.000000, 0.000000, 0.0000...\nllama_model_loader: - kv 16: tokenizer.ggml.token_type arr[i32,131072] = [3, 3, 3, 2, 1, 1, 1, 1, 1, 1, 1, 1, ...\nllama_model_loader: - kv 17: tokenizer.ggml.bos_token_id u32 = 1\nllama_model_loader: - kv 18: tokenizer.ggml.eos_token_id u32 = 2\nllama_model_loader: - kv 19: tokenizer.ggml.unknown_token_id u32 = 3\nllama_model_loader: - kv 20: tokenizer.ggml.padding_token_id u32 = 0\nllama_model_loader: - kv 21: tokenizer.ggml.add_bos_token bool = false\nllama_model_loader: - kv 22: tokenizer.ggml.add_eos_token bool = false\nllama_model_loader: - kv 23: general.quantization_version u32 = 2\nllama_model_loader: - kv 24: split.no u16 = 0\nllama_model_loader: - kv 25: split.count u16 = 9\nllama_model_loader: - kv 26: split.tensors.count i32 = 2114\nllama_model_loader: - type f32: 257 tensors\nllama_model_loader: - type f16: 64 tensors\nllama_model_loader: - type q8_0: 128 tensors\nllama_model_loader: - type q2_K: 1088 tensors\nllama_model_loader: - type q3_K: 512 tensors\nllama_model_loader: - type q5_K: 64 tensors\nllama_model_loader: - type q6_K: 1 tensors\nllm_load_vocab: mismatch in special tokens definition ( 284\/131072 vs 260\/131072 ).\nllm_load_print_meta: format = GGUF V3 (latest)\nllm_load_print_meta: arch = grok\nllm_load_print_meta: vocab type = SPM\nllm_load_print_meta: n_vocab = 131072\nllm_load_print_meta: n_merges = 0\nllm_load_print_meta: n_ctx_train = 4096\nllm_load_print_meta: n_embd = 6144\nllm_load_print_meta: n_head = 48\nllm_load_print_meta: n_head_kv = 8\nllm_load_print_meta: n_layer = 64\nllm_load_print_meta: n_rot = 128\nllm_load_print_meta: n_embd_head_k = 128\nllm_load_print_meta: n_embd_head_v = 128\nllm_load_print_meta: n_gqa = 6\nllm_load_print_meta: n_embd_k_gqa = 1024\nllm_load_print_meta: n_embd_v_gqa = 1024\nllm_load_print_meta: f_norm_eps = 0.0e+00\nllm_load_print_meta: f_norm_rms_eps = 1.0e-05\nllm_load_print_meta: f_clamp_kqv = 0.0e+00\nllm_load_print_meta: f_max_alibi_bias = 0.0e+00\nllm_load_print_meta: f_logit_scale = 0.0e+00\nllm_load_print_meta: n_ff = 32768\nllm_load_print_meta: n_expert = 8\nllm_load_print_meta: n_expert_used = 2\nllm_load_print_meta: causal attn = 1\nllm_load_print_meta: pooling type = 0\nllm_load_print_meta: rope type = 2\nllm_load_print_meta: rope scaling = linear\nllm_load_print_meta: freq_base_train = 100000.0\nllm_load_print_meta: freq_scale_train = 1\nllm_load_print_meta: n_yarn_orig_ctx = 4096\nllm_load_print_meta: rope_finetuned = unknown\nllm_load_print_meta: ssm_d_conv = 0\nllm_load_print_meta: ssm_d_inner = 0\nllm_load_print_meta: ssm_d_state = 0\nllm_load_print_meta: ssm_dt_rank = 0\nllm_load_print_meta: model type = 314B\nllm_load_print_meta: model ftype = Q2_K - Medium\nllm_load_print_meta: model params = 315.68 B\nllm_load_print_meta: model size = 107.96 GiB (2.94 BPW)\nllm_load_print_meta: general.name = Grok\nllm_load_print_meta: BOS token = 1 '[BOS]'\nllm_load_print_meta: EOS token = 2 '[EOS]'\nllm_load_print_meta: UNK token = 3 '[UNK]'\nllm_load_print_meta: PAD token = 0 '[PAD]'\nllm_load_print_meta: LF token = 79 '<0x0A>'\nggml_cuda_init: GGML_CUDA_FORCE_MMQ: no\nggml_cuda_init: CUDA_USE_TENSOR_CORES: yes\nggml_cuda_init: found 8 CUDA devices:\n Device 0: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\n Device 1: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\n Device 2: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\n Device 3: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\n Device 4: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\n Device 5: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\n Device 6: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\n Device 7: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes\nllm_load_tensors: ggml ctx size = 7.26 MiB\nllm_load_tensors: offloading 64 repeating layers to GPU\nllm_load_tensors: offloading non-repeating layers to GPU\nllm_load_tensors: offloaded 65\/65 layers to GPU\nllm_load_tensors: CPU buffer size = 630.00 MiB\nllm_load_tensors: CUDA0 buffer size = 15457.50 MiB\nllm_load_tensors: CUDA1 buffer size = 13740.00 MiB\nllm_load_tensors: CUDA2 buffer size = 13740.00 MiB\nllm_load_tensors: CUDA3 buffer size = 13740.00 MiB\nllm_load_tensors: CUDA4 buffer size = 13740.00 MiB\nllm_load_tensors: CUDA5 buffer size = 13740.00 MiB\nllm_load_tensors: CUDA6 buffer size = 13740.00 MiB\nllm_load_tensors: CUDA7 buffer size = 12652.52 MiB\n....................................................................................................\nllama_new_context_with_model: n_ctx = 4096\nllama_new_context_with_model: n_batch = 2048\nllama_new_context_with_model: n_ubatch = 512\nllama_new_context_with_model: freq_base = 100000.0\nllama_new_context_with_model: freq_scale = 1\nllama_kv_cache_init: CUDA0 KV buffer size = 144.00 MiB\nllama_kv_cache_init: CUDA1 KV buffer size = 128.00 MiB\nllama_kv_cache_init: CUDA2 KV buffer size = 128.00 MiB\nllama_kv_cache_init: CUDA3 KV buffer size = 128.00 MiB\nllama_kv_cache_init: CUDA4 KV buffer size = 128.00 MiB\nllama_kv_cache_init: CUDA5 KV buffer size = 128.00 MiB\nllama_kv_cache_init: CUDA6 KV buffer size = 128.00 MiB\nllama_kv_cache_init: CUDA7 KV buffer size = 112.00 MiB\nllama_new_context_with_model: KV self size = 1024.00 MiB, K (f16): 512.00 MiB, V (f16): 512.00 MiB\nllama_new_context_with_model: CUDA_Host output buffer size = 0.50 MiB\nllama_new_context_with_model: pipeline parallelism enabled (n_copies=4)\nllama_new_context_with_model: CUDA0 compute buffer size = 564.04 MiB\nllama_new_context_with_model: CUDA1 compute buffer size = 564.04 MiB\nllama_new_context_with_model: CUDA2 compute buffer size = 564.04 MiB\nllama_new_context_with_model: CUDA3 compute buffer size = 564.04 MiB\nllama_new_context_with_model: CUDA4 compute buffer size = 564.04 MiB\nllama_new_context_with_model: CUDA5 compute buffer size = 564.04 MiB\nllama_new_context_with_model: CUDA6 compute buffer size = 564.04 MiB\nllama_new_context_with_model: CUDA7 compute buffer size = 564.05 MiB\nllama_new_context_with_model: CUDA_Host compute buffer size = 44.02 MiB\nllama_new_context_with_model: graph nodes = 3784\nllama_new_context_with_model: graph splits = 9\n\nsystem_info: n_threads = 16 \/ 64 | AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 1 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 0 | VSX = 0 | MATMUL_INT8 = 0 |\nsampling:\n repeat_last_n = 64, repeat_penalty = 1.100, frequency_penalty = 0.000, presence_penalty = 0.000\n top_k = 40, tfs_z = 1.000, top_p = 0.950, min_p = 0.050, typical_p = 1.000, temp = 0.900\n mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000\nsampling order:\nCFG -> Penalties -> top_k -> tfs_z -> typical_p -> top_p -> min_p -> temperature\ngenerate: n_ctx = 4096, n_batch = 2048, n_predict = -1, n_keep = 0\n\n\n ### Instruction: Write me a linked list implementation in C\/C++, starting with '```' and ending with '```'\\n### Response:\n\n<details><summary>Click to expand<\/summary>\n<p>\n#include <stdio.h>\n#include <stdlib.h>\n\nstruct node {\n int data;\n struct node *next;\n};\n\nvoid push(struct node **head, int value) {\n struct node *node = malloc(sizeof(struct node));\n node->data = value;\n node->next = *head;\n *head = node;\n}\n\nint main() {\n \/\/ Create a new head pointer for the list.\n struct node **list = (struct node **) calloc(1, sizeof(struct node *));\n push(&(*list), 0); \/\/ Insert a value into the list.\n for (int i = 1; i < 8; i++) {\n \/\/ Create new nodes for each element in array and add to linked-list.\n if (!i) continue;\n push(list, i);\n }\n\n struct node *current = *list;\n while (current->next) {\n printf(\"%d \", current->data);\n current = (*list)->next;\n }\n printf(\"\\n\"); \/\/ 1234567\n}\n\n<\/p>\n<\/details> [end of text]\n\nllama_print_timings: load time = 142096.82 ms\nllama_print_timings: sample time = 174.00 ms \/ 281 runs ( 0.62 ms per token, 1614.90 tokens per second)\nllama_print_timings: prompt eval time = 2152.91 ms \/ 29 tokens ( 74.24 ms per token, 13.47 tokens per second)\nllama_print_timings: eval time = 257631.30 ms \/ 280 runs ( 920.11 ms per token, 1.09 tokens per second)\nllama_print_timings: total time = 260199.48 ms \/ 309 tokens\nLog end\n<\/pre><\/div>\n","protected":false},"excerpt":{"rendered":"\u6d4b\u8bd5\u73af\u5883 windows, llama.cpp, cuda, \u8fd0\u884c\u547d\u4ee4\uff1a<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[313,289,443,442,312],"tags":[398],"class_list":["post-2619","post","type-post","status-publish","format-standard","hentry","category-chatgpt","category-gpt","category-llm","category-llms","category-openai","tag-grok-1"],"views":1932,"jetpack_sharing_enabled":true,"jetpack_featured_media_url":"","_links":{"self":[{"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=\/wp\/v2\/posts\/2619","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2619"}],"version-history":[{"count":4,"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=\/wp\/v2\/posts\/2619\/revisions"}],"predecessor-version":[{"id":2623,"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=\/wp\/v2\/posts\/2619\/revisions\/2623"}],"wp:attachment":[{"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2619"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2619"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.aqwu.net\/wp\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2619"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}