llama_model_loader: loaded meta data with 29 key-value pairs and 254 tensors from RoGemma-7b-Instruct-IMat-GGUF/RoGemma-7b-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 = gemma llama_model_loader: - kv 1: general.name str = RoGemma-7b-Instruct llama_model_loader: - kv 2: gemma.context_length u32 = 8192 llama_model_loader: - kv 3: gemma.embedding_length u32 = 3072 llama_model_loader: - kv 4: gemma.block_count u32 = 28 llama_model_loader: - kv 5: gemma.feed_forward_length u32 = 24576 llama_model_loader: - kv 6: gemma.attention.head_count u32 = 16 llama_model_loader: - kv 7: gemma.attention.head_count_kv u32 = 16 llama_model_loader: - kv 8: gemma.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 9: gemma.attention.key_length u32 = 256 llama_model_loader: - kv 10: gemma.attention.value_length u32 = 256 llama_model_loader: - kv 11: general.file_type u32 = 7 llama_model_loader: - kv 12: tokenizer.ggml.model str = llama llama_model_loader: - kv 13: tokenizer.ggml.pre str = default llama_model_loader: - kv 14: tokenizer.ggml.tokens arr[str,256000] = ["", "", "", "", ... llama_model_loader: - kv 15: tokenizer.ggml.scores arr[f32,256000] = [0.000000, 0.000000, 0.000000, 0.0000... llama_model_loader: - kv 16: tokenizer.ggml.token_type arr[i32,256000] = [3, 3, 3, 2, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 17: tokenizer.ggml.bos_token_id u32 = 2 llama_model_loader: - kv 18: tokenizer.ggml.eos_token_id u32 = 1 llama_model_loader: - kv 19: tokenizer.ggml.unknown_token_id u32 = 3 llama_model_loader: - kv 20: tokenizer.ggml.padding_token_id u32 = 0 llama_model_loader: - kv 21: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 22: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 23: tokenizer.chat_template str = {{ '' }}{% if messages[0]['role'... llama_model_loader: - kv 24: tokenizer.ggml.prefix_token_id u32 = 67 llama_model_loader: - kv 25: tokenizer.ggml.suffix_token_id u32 = 69 llama_model_loader: - kv 26: tokenizer.ggml.middle_token_id u32 = 68 llama_model_loader: - kv 27: tokenizer.ggml.eot_token_id u32 = 107 llama_model_loader: - kv 28: general.quantization_version u32 = 2 llama_model_loader: - type f32: 57 tensors llama_model_loader: - type q8_0: 197 tensors llm_load_vocab: special tokens cache size = 260 llm_load_vocab: token to piece cache size = 1.6014 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = gemma llm_load_print_meta: vocab type = SPM llm_load_print_meta: n_vocab = 256000 llm_load_print_meta: n_merges = 0 llm_load_print_meta: n_ctx_train = 8192 llm_load_print_meta: n_embd = 3072 llm_load_print_meta: n_head = 16 llm_load_print_meta: n_head_kv = 16 llm_load_print_meta: n_layer = 28 llm_load_print_meta: n_rot = 192 llm_load_print_meta: n_embd_head_k = 256 llm_load_print_meta: n_embd_head_v = 256 llm_load_print_meta: n_gqa = 1 llm_load_print_meta: n_embd_k_gqa = 4096 llm_load_print_meta: n_embd_v_gqa = 4096 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-06 llm_load_print_meta: f_clamp_kqv = 0.0e+00 llm_load_print_meta: f_max_alibi_bias = 0.0e+00 llm_load_print_meta: f_logit_scale = 0.0e+00 llm_load_print_meta: n_ff = 24576 llm_load_print_meta: n_expert = 0 llm_load_print_meta: n_expert_used = 0 llm_load_print_meta: causal attn = 1 llm_load_print_meta: pooling type = 0 llm_load_print_meta: rope type = 2 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 10000.0 llm_load_print_meta: freq_scale_train = 1 llm_load_print_meta: n_ctx_orig_yarn = 8192 llm_load_print_meta: rope_finetuned = unknown llm_load_print_meta: ssm_d_conv = 0 llm_load_print_meta: ssm_d_inner = 0 llm_load_print_meta: ssm_d_state = 0 llm_load_print_meta: ssm_dt_rank = 0 llm_load_print_meta: model type = 7B llm_load_print_meta: model ftype = Q8_0 llm_load_print_meta: model params = 8.54 B llm_load_print_meta: model size = 8.45 GiB (8.50 BPW) llm_load_print_meta: general.name = RoGemma-7b-Instruct llm_load_print_meta: BOS token = 2 '' llm_load_print_meta: EOS token = 1 '' llm_load_print_meta: UNK token = 3 '' llm_load_print_meta: PAD token = 0 '' llm_load_print_meta: LF token = 227 '<0x0A>' llm_load_print_meta: PRE token = 67 '' llm_load_print_meta: SUF token = 69 '' llm_load_print_meta: MID token = 68 '' llm_load_print_meta: EOT token = 107 '' llm_load_print_meta: max token length = 93 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 = 0.24 MiB llm_load_tensors: offloading 28 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU llm_load_tensors: offloaded 29/29 layers to GPU llm_load_tensors: CPU buffer size = 796.88 MiB llm_load_tensors: CUDA0 buffer size = 8651.54 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 = 224.00 MiB llama_new_context_with_model: KV self size = 224.00 MiB, K (f16): 112.00 MiB, V (f16): 112.00 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.98 MiB llama_new_context_with_model: CUDA0 compute buffer size = 506.00 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 7.01 MiB llama_new_context_with_model: graph nodes = 931 llama_new_context_with_model: graph splits = 2 system_info: n_threads = 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 122.982 ms compute_imatrix: computing over 128 chunks with batch_size 512 compute_imatrix: 0.72 seconds per pass - ETA 1.53 minutes [1]6.7268,[2]4.7956,[3]4.3365,[4]5.4634,[5]5.5761,[6]4.7844,[7]5.2281,[8]5.4576,[9]5.6880, save_imatrix: stored collected data after 10 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [10]5.1084,[11]5.2387,[12]5.6425,[13]6.0984,[14]6.4094,[15]6.7641,[16]7.0334,[17]7.1327,[18]7.4186,[19]7.1324, save_imatrix: stored collected data after 20 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [20]7.2420,[21]7.3956,[22]7.3754,[23]7.4916,[24]7.5334,[25]7.6872,[26]7.4682,[27]7.7364,[28]7.9965,[29]7.9577, save_imatrix: stored collected data after 30 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [30]7.9155,[31]7.4608,[32]7.2029,[33]7.1080,[34]6.9745,[35]6.9006,[36]7.1893,[37]7.2251,[38]7.2493,[39]7.3769, save_imatrix: stored collected data after 40 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [40]7.4966,[41]7.6572,[42]7.9743,[43]8.2865,[44]8.5892,[45]8.7742,[46]8.6579,[47]8.6807,[48]8.8624,[49]8.9978, save_imatrix: stored collected data after 50 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [50]8.8560,[51]8.8432,[52]8.8797,[53]8.9995,[54]9.1008,[55]9.2532,[56]9.3016,[57]9.3064,[58]9.3182,[59]9.1306, save_imatrix: stored collected data after 60 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [60]9.0207,[61]8.8842,[62]8.8355,[63]8.8759,[64]8.8721,[65]8.8531,[66]8.8684,[67]8.8022,[68]8.7356,[69]8.7641, save_imatrix: stored collected data after 70 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [70]8.7371,[71]8.7328,[72]8.7399,[73]8.7070,[74]8.6636,[75]8.6279,[76]8.6341,[77]8.6556,[78]8.6528,[79]8.6093, save_imatrix: stored collected data after 80 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [80]8.6513,[81]8.6853,[82]8.6525,[83]8.6525,[84]8.6889,[85]8.5525,[86]8.5107,[87]8.4442,[88]8.4460,[89]8.4832, save_imatrix: stored collected data after 90 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [90]8.4874,[91]8.4152,[92]8.3457,[93]8.2611,[94]8.1827,[95]8.1159,[96]8.0418,[97]7.9787,[98]7.9182,[99]7.9335, save_imatrix: stored collected data after 100 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [100]7.9383,[101]8.0119,[102]8.0867,[103]8.1573,[104]8.2901,[105]8.3851,[106]8.4142,[107]8.4310,[108]8.4466,[109]8.4205, save_imatrix: stored collected data after 110 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [110]8.4063,[111]8.3474,[112]8.2734,[113]8.3102,[114]8.3227,[115]8.3146,[116]8.3066,[117]8.3351,[118]8.3537,[119]8.3591, save_imatrix: stored collected data after 120 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat [120]8.3618,[121]8.3700,[122]8.3300,[123]8.3984,[124]8.4737,[125]8.5299,[126]8.6142,[127]8.6792,[128]8.7451, save_imatrix: stored collected data after 128 chunks in RoGemma-7b-Instruct-IMat-GGUF/imatrix.dat llama_print_timings: load time = 6235.43 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 = 76873.60 ms / 65536 tokens ( 1.17 ms per token, 852.52 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 = 83878.25 ms / 65537 tokens Final estimate: PPL = 8.7451 +/- 0.12878