llama_model_loader: loaded meta data with 34 key-value pairs and 508 tensors from shieldgemma-27b-IMat-GGUF/shieldgemma-27b.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 = gemma2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Shieldgemma 27b llama_model_loader: - kv 3: general.basename str = shieldgemma llama_model_loader: - kv 4: general.size_label str = 27B llama_model_loader: - kv 5: general.license str = gemma llama_model_loader: - kv 6: general.tags arr[str,1] = ["text-generation"] llama_model_loader: - kv 7: gemma2.context_length u32 = 8192 llama_model_loader: - kv 8: gemma2.embedding_length u32 = 4608 llama_model_loader: - kv 9: gemma2.block_count u32 = 46 llama_model_loader: - kv 10: gemma2.feed_forward_length u32 = 36864 llama_model_loader: - kv 11: gemma2.attention.head_count u32 = 32 llama_model_loader: - kv 12: gemma2.attention.head_count_kv u32 = 16 llama_model_loader: - kv 13: gemma2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 14: gemma2.attention.key_length u32 = 128 llama_model_loader: - kv 15: gemma2.attention.value_length u32 = 128 llama_model_loader: - kv 16: general.file_type u32 = 7 llama_model_loader: - kv 17: gemma2.attn_logit_softcapping f32 = 50.000000 llama_model_loader: - kv 18: gemma2.final_logit_softcapping f32 = 30.000000 llama_model_loader: - kv 19: gemma2.attention.sliding_window u32 = 4096 llama_model_loader: - kv 20: tokenizer.ggml.model str = llama llama_model_loader: - kv 21: tokenizer.ggml.pre str = default llama_model_loader: - kv 22: tokenizer.ggml.tokens arr[str,256000] = ["", "", "", "", ... llama_model_loader: - kv 23: tokenizer.ggml.scores arr[f32,256000] = [-1000.000000, -1000.000000, -1000.00... llama_model_loader: - kv 24: tokenizer.ggml.token_type arr[i32,256000] = [3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, ... llama_model_loader: - kv 25: tokenizer.ggml.bos_token_id u32 = 2 llama_model_loader: - kv 26: tokenizer.ggml.eos_token_id u32 = 1 llama_model_loader: - kv 27: tokenizer.ggml.unknown_token_id u32 = 3 llama_model_loader: - kv 28: tokenizer.ggml.padding_token_id u32 = 0 llama_model_loader: - kv 29: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 30: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 31: tokenizer.chat_template str = {{- bos_token }}\n{%- if messages[-1].... llama_model_loader: - kv 32: tokenizer.ggml.add_space_prefix bool = false llama_model_loader: - kv 33: general.quantization_version u32 = 2 llama_model_loader: - type f32: 185 tensors llama_model_loader: - type q8_0: 323 tensors llm_load_vocab: special tokens cache size = 249 llm_load_vocab: token to piece cache size = 1.6014 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = gemma2 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: vocab_only = 0 llm_load_print_meta: n_ctx_train = 8192 llm_load_print_meta: n_embd = 4608 llm_load_print_meta: n_layer = 46 llm_load_print_meta: n_head = 32 llm_load_print_meta: n_head_kv = 16 llm_load_print_meta: n_rot = 128 llm_load_print_meta: n_swa = 4096 llm_load_print_meta: n_embd_head_k = 128 llm_load_print_meta: n_embd_head_v = 128 llm_load_print_meta: n_gqa = 2 llm_load_print_meta: n_embd_k_gqa = 2048 llm_load_print_meta: n_embd_v_gqa = 2048 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 = 36864 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 = 27B llm_load_print_meta: model ftype = Q8_0 llm_load_print_meta: model params = 27.23 B llm_load_print_meta: model size = 26.94 GiB (8.50 BPW) llm_load_print_meta: general.name = Shieldgemma 27b 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: EOT token = 107 '' llm_load_print_meta: max token length = 48 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 = 0.45 MiB llm_load_tensors: offloading 37 repeating layers to GPU llm_load_tensors: offloaded 37/47 layers to GPU llm_load_tensors: CPU buffer size = 27591.06 MiB llm_load_tensors: CUDA0 buffer size = 21231.35 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: CUDA_Host KV buffer size = 36.00 MiB llama_kv_cache_init: CUDA0 KV buffer size = 148.00 MiB llama_new_context_with_model: KV self size = 184.00 MiB, K (f16): 92.00 MiB, V (f16): 92.00 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.98 MiB llama_new_context_with_model: CUDA0 compute buffer size = 1704.31 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 11.01 MiB llama_new_context_with_model: graph nodes = 1850 llama_new_context_with_model: graph splits = 121 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 127.59 ms compute_imatrix: computing over 128 chunks with batch_size 512 compute_imatrix: 2.04 seconds per pass - ETA 4.35 minutes [1]6.5529,[2]4.3753,[3]3.8200,[4]4.7426,[5]4.8615,[6]4.2138,[7]4.5153,[8]4.7561,[9]5.0004, save_imatrix: stored collected data after 10 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [10]4.4829,[11]4.6406,[12]5.0627,[13]5.5336,[14]5.7434,[15]6.1916,[16]6.4807,[17]6.6260,[18]6.9112,[19]6.6039, save_imatrix: stored collected data after 20 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [20]6.7241,[21]6.8570,[22]6.8188,[23]6.9620,[24]7.0188,[25]7.1975,[26]6.9847,[27]7.1626,[28]7.2851,[29]7.2054, save_imatrix: stored collected data after 30 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [30]7.2006,[31]6.7962,[32]6.5772,[33]6.4917,[34]6.3837,[35]6.3487,[36]6.3662,[37]6.3802,[38]6.4268,[39]6.5645, save_imatrix: stored collected data after 40 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [40]6.6992,[41]6.8132,[42]7.0489,[43]7.2844,[44]7.5078,[45]7.6526,[46]7.5486,[47]7.5756,[48]7.7402,[49]7.8650, save_imatrix: stored collected data after 50 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [50]7.7079,[51]7.7449,[52]7.7805,[53]7.8870,[54]8.0256,[55]8.1275,[56]8.1925,[57]8.2110,[58]8.2431,[59]8.1083, save_imatrix: stored collected data after 60 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [60]8.0200,[61]7.9049,[62]7.8772,[63]7.9039,[64]7.9035,[65]7.8704,[66]7.8759,[67]7.8261,[68]7.7754,[69]7.7932, save_imatrix: stored collected data after 70 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [70]7.7601,[71]7.7481,[72]7.7565,[73]7.7407,[74]7.6986,[75]7.6610,[76]7.6540,[77]7.6525,[78]7.6471,[79]7.6015, save_imatrix: stored collected data after 80 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [80]7.6499,[81]7.6832,[82]7.6675,[83]7.6768,[84]7.7121,[85]7.6041,[86]7.5667,[87]7.5087,[88]7.5209,[89]7.5609, save_imatrix: stored collected data after 90 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [90]7.5806,[91]7.5265,[92]7.4684,[93]7.3988,[94]7.3356,[95]7.2910,[96]7.2315,[97]7.1779,[98]7.1320,[99]7.1597, save_imatrix: stored collected data after 100 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [100]7.1820,[101]7.2535,[102]7.3217,[103]7.3924,[104]7.5169,[105]7.6095,[106]7.6266,[107]7.6487,[108]7.6654,[109]7.6394, save_imatrix: stored collected data after 110 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [110]7.6112,[111]7.5207,[112]7.4226,[113]7.4681,[114]7.4921,[115]7.4919,[116]7.4894,[117]7.5252,[118]7.5394,[119]7.5509, save_imatrix: stored collected data after 120 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat [120]7.5613,[121]7.5822,[122]7.5593,[123]7.6007,[124]7.6477,[125]7.6850,[126]7.7480,[127]7.7968,[128]7.8351, save_imatrix: stored collected data after 128 chunks in shieldgemma-27b-IMat-GGUF/imatrix.dat llama_print_timings: load time = 4902.13 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 = 227177.99 ms / 65536 tokens ( 3.47 ms per token, 288.48 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 = 231584.30 ms / 65537 tokens Final estimate: PPL = 7.8351 +/- 0.11938