llama_model_loader: loaded meta data with 32 key-value pairs and 628 tensors from granite-20b-code-instruct-IMat-GGUF/granite-20b-code-instruct.Q8_0.gguf.hardlink.gguf (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = starcoder llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Granite 20b Code Instruct llama_model_loader: - kv 3: general.finetune str = code-instruct llama_model_loader: - kv 4: general.basename str = granite llama_model_loader: - kv 5: general.size_label str = 20B llama_model_loader: - kv 6: general.license str = apache-2.0 llama_model_loader: - kv 7: general.base_model.count u32 = 1 llama_model_loader: - kv 8: general.base_model.0.name str = Granite 20b Code Base llama_model_loader: - kv 9: general.base_model.0.organization str = Ibm Granite llama_model_loader: - kv 10: general.base_model.0.repo_url str = https://huggingface.co/ibm-granite/gr... llama_model_loader: - kv 11: general.tags arr[str,3] = ["code", "granite", "text-generation"] llama_model_loader: - kv 12: general.datasets arr[str,8] = ["bigcode/commitpackft", "TIGER-Lab/M... llama_model_loader: - kv 13: starcoder.context_length u32 = 8192 llama_model_loader: - kv 14: starcoder.embedding_length u32 = 6144 llama_model_loader: - kv 15: starcoder.feed_forward_length u32 = 24576 llama_model_loader: - kv 16: starcoder.block_count u32 = 52 llama_model_loader: - kv 17: starcoder.attention.head_count u32 = 48 llama_model_loader: - kv 18: starcoder.attention.head_count_kv u32 = 1 llama_model_loader: - kv 19: starcoder.attention.layer_norm_epsilon f32 = 0.000010 llama_model_loader: - kv 20: general.file_type u32 = 7 llama_model_loader: - kv 21: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 22: tokenizer.ggml.pre str = refact llama_model_loader: - kv 23: tokenizer.ggml.tokens arr[str,49152] = ["<|endoftext|>", "", "' llm_load_print_meta: EOS token = 0 '<|endoftext|>' llm_load_print_meta: UNK token = 0 '<|endoftext|>' llm_load_print_meta: PAD token = 0 '<|endoftext|>' llm_load_print_meta: LF token = 145 'Ä' llm_load_print_meta: EOT token = 0 '<|endoftext|>' llm_load_print_meta: max token length = 512 ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes llm_load_tensors: ggml ctx size = 0.55 MiB llm_load_tensors: offloading 52 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU llm_load_tensors: offloaded 53/53 layers to GPU llm_load_tensors: CPU buffer size = 498.00 MiB llm_load_tensors: CUDA0 buffer size = 20292.38 MiB .................................................................................................... llama_new_context_with_model: n_ctx = 512 llama_new_context_with_model: n_batch = 512 llama_new_context_with_model: n_ubatch = 512 llama_new_context_with_model: flash_attn = 0 llama_new_context_with_model: freq_base = 10000.0 llama_new_context_with_model: freq_scale = 1 llama_kv_cache_init: CUDA0 KV buffer size = 13.00 MiB llama_new_context_with_model: KV self size = 13.00 MiB, K (f16): 6.50 MiB, V (f16): 6.50 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.19 MiB llama_new_context_with_model: CUDA0 compute buffer size = 120.00 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 25.01 MiB llama_new_context_with_model: graph nodes = 1933 llama_new_context_with_model: graph splits = 2 system_info: n_threads = 25 (n_threads_batch = 25) / 32 | AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 1 | AVX512_VBMI = 1 | AVX512_VNNI = 1 | AVX512_BF16 = 1 | FMA = 1 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | compute_imatrix: tokenizing the input .. compute_imatrix: tokenization took 113.128 ms compute_imatrix: computing over 152 chunks with batch_size 512 compute_imatrix: 1.30 seconds per pass - ETA 3.28 minutes [1]5.5316,[2]4.3038,[3]4.5649,[4]5.0288,[5]5.6039,[6]5.7152,[7]5.0016,[8]5.9597,[9]5.9637, save_imatrix: stored collected data after 10 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [10]6.7819,[11]7.0343,[12]6.4546,[13]7.0672,[14]7.6034,[15]8.3912,[16]8.5493,[17]9.1657,[18]9.4702,[19]9.7523, save_imatrix: stored collected data after 20 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [20]9.9486,[21]10.4114,[22]10.0740,[23]10.0213,[24]10.3530,[25]10.5995,[26]10.6331,[27]10.4010,[28]10.5028,[29]10.7926, save_imatrix: stored collected data after 30 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [30]11.0959,[31]11.0268,[32]11.1389,[33]11.3398,[34]11.6580,[35]11.7255,[36]11.4972,[37]10.8076,[38]10.3152,[39]10.2085, save_imatrix: stored collected data after 40 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [40]10.1415,[41]10.0520,[42]10.1053,[43]9.9861,[44]10.0245,[45]10.0257,[46]10.1471,[47]10.1358,[48]10.3092,[49]10.5322, save_imatrix: stored collected data after 50 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [50]10.4987,[51]10.8835,[52]11.2237,[53]11.6686,[54]11.9040,[55]12.2365,[56]12.2093,[57]12.1756,[58]12.1845,[59]12.3144, save_imatrix: stored collected data after 60 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [60]12.3865,[61]12.2209,[62]12.2149,[63]12.2819,[64]12.3703,[65]12.4174,[66]12.4922,[67]12.6439,[68]12.7639,[69]12.7716, save_imatrix: stored collected data after 70 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [70]12.8490,[71]12.8158,[72]12.6971,[73]12.6002,[74]12.4168,[75]12.3231,[76]12.4371,[77]12.5235,[78]12.4813,[79]12.4734, save_imatrix: stored collected data after 80 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [80]12.4798,[81]12.3699,[82]12.2741,[83]12.2127,[84]12.1877,[85]12.1598,[86]12.0773,[87]12.0180,[88]11.9234,[89]11.8501, save_imatrix: stored collected data after 90 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [90]11.7759,[91]11.7479,[92]11.6583,[93]11.5726,[94]11.4962,[95]11.4289,[96]11.4031,[97]11.3468,[98]11.3321,[99]11.2733, save_imatrix: stored collected data after 100 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [100]11.2175,[101]11.2073,[102]11.1154,[103]11.0428,[104]11.0297,[105]11.0836,[106]11.1179,[107]11.1815,[108]11.2633,[109]11.1341, save_imatrix: stored collected data after 110 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [110]11.0276,[111]10.9195,[112]10.7971,[113]10.6767,[114]10.5720,[115]10.4712,[116]10.3728,[117]10.3827,[118]10.3708,[119]10.3970, save_imatrix: stored collected data after 120 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [120]10.5022,[121]10.6177,[122]10.7352,[123]10.7968,[124]10.9187,[125]10.9587,[126]10.9877,[127]11.0106,[128]11.0613,[129]11.0879, save_imatrix: stored collected data after 130 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [130]11.0612,[131]11.0532,[132]11.0083,[133]11.0206,[134]11.0336,[135]11.0436,[136]11.0618,[137]11.0628,[138]11.0782,[139]11.0803, save_imatrix: stored collected data after 140 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [140]11.1238,[141]11.1233,[142]11.1360,[143]11.1245,[144]11.0680,[145]11.0480,[146]11.1301,[147]11.2386,[148]11.3118,[149]11.4027, save_imatrix: stored collected data after 150 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat [150]11.4866,[151]11.5899,[152]11.6361, save_imatrix: stored collected data after 152 chunks in granite-20b-code-instruct-IMat-GGUF/imatrix.dat llama_print_timings: load time = 11508.22 ms llama_print_timings: sample time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second) llama_print_timings: prompt eval time = 173482.28 ms / 77824 tokens ( 2.23 ms per token, 448.60 tokens per second) llama_print_timings: eval time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second) llama_print_timings: total time = 184251.37 ms / 77825 tokens Final estimate: PPL = 11.6361 +/- 0.19854