main: build = 3051 (5921b8f0) main: built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu main: seed = 1717152257 llama_model_loader: loaded meta data with 24 key-value pairs and 255 tensors from neo_7b_sft_v0.1-IMat-GGUF/neo_7b_sft_v0.1.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 = llama llama_model_loader: - kv 1: general.name str = neo_7b_sft_v0.1 llama_model_loader: - kv 2: llama.block_count u32 = 28 llama_model_loader: - kv 3: llama.context_length u32 = 8192 llama_model_loader: - kv 4: llama.embedding_length u32 = 3072 llama_model_loader: - kv 5: llama.feed_forward_length u32 = 24576 llama_model_loader: - kv 6: llama.attention.head_count u32 = 16 llama_model_loader: - kv 7: llama.attention.head_count_kv u32 = 16 llama_model_loader: - kv 8: llama.rope.freq_base f32 = 10000.000000 llama_model_loader: - kv 9: llama.attention.layer_norm_rms_epsilon f32 = 0.000010 llama_model_loader: - kv 10: general.file_type u32 = 0 llama_model_loader: - kv 11: llama.vocab_size u32 = 64256 llama_model_loader: - kv 12: llama.rope.dimension_count u32 = 192 llama_model_loader: - kv 13: tokenizer.ggml.model str = llama llama_model_loader: - kv 14: tokenizer.ggml.pre str = default llama_model_loader: - kv 15: tokenizer.ggml.tokens arr[str,64256] = ["", "~~", "~~", "<0x00>", "<... llama_model_loader: - kv 16: tokenizer.ggml.scores arr[f32,64256] = [0.000000, 0.000000, 0.000000, 0.0000... llama_model_loader: - kv 17: tokenizer.ggml.token_type arr[i32,64256] = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ... llama_model_loader: - kv 18: tokenizer.ggml.bos_token_id u32 = 1 llama_model_loader: - kv 19: tokenizer.ggml.eos_token_id u32 = 2 llama_model_loader: - kv 20: tokenizer.ggml.add_bos_token bool = false llama_model_loader: - kv 21: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 22: tokenizer.chat_template str = {% set system_message = 'You are a he... llama_model_loader: - kv 23: general.quantization_version u32 = 2 llama_model_loader: - type f32: 255 tensors llm_load_vocab: special tokens cache size = 515 llm_load_vocab: token to piece cache size = 0.7263 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = llama llm_load_print_meta: vocab type = SPM llm_load_print_meta: n_vocab = 64256 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 = 192 llm_load_print_meta: n_embd_head_v = 192 llm_load_print_meta: n_gqa = 1 llm_load_print_meta: n_embd_k_gqa = 3072 llm_load_print_meta: n_embd_v_gqa = 3072 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-05 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 = 0 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_yarn_orig_ctx = 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 = ?B llm_load_print_meta: model ftype = all F32 llm_load_print_meta: model params = 7.79 B llm_load_print_meta: model size = 29.03 GiB (32.00 BPW) llm_load_print_meta: general.name = neo_7b_sft_v0.1 llm_load_print_meta: BOS token = 1 '~~' llm_load_print_meta: EOS token = 2 '~~' llm_load_print_meta: UNK token = 0 '' llm_load_print_meta: LF token = 13 '<0x0A>' ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes 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 = 0.26 MiB llm_load_tensors: offloading 22 repeating layers to GPU llm_load_tensors: offloaded 22/29 layers to GPU llm_load_tensors: CPU buffer size = 29730.67 MiB llm_load_tensors: CUDA0 buffer size = 22176.52 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 = 132.00 MiB llama_new_context_with_model: KV self size = 168.00 MiB, K (f16): 84.00 MiB, V (f16): 84.00 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.25 MiB llama_new_context_with_model: CUDA0 compute buffer size = 884.50 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 13.01 MiB llama_new_context_with_model: graph nodes = 902 llama_new_context_with_model: graph splits = 70 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 94.53 ms compute_imatrix: computing over 149 chunks with batch_size 512 compute_imatrix: 0.92 seconds per pass - ETA 2.28 minutes [1]6.9087,[2]4.9464,[3]5.5776,[4]5.6370,[5]5.4654,[6]5.4102,[7]4.7098,[8]4.8687,[9]4.9457, save_imatrix: stored collected data after 10 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [10]5.3631,[11]5.3031,[12]4.9517,[13]5.3067,[14]5.7762,[15]6.0604,[16]6.5393,[17]6.8275,[18]7.0775,[19]7.2396, save_imatrix: stored collected data after 20 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [20]7.6090,[21]7.3325,[22]7.3046,[23]7.4432,[24]7.6053,[25]7.5198,[26]7.4926,[27]7.6715,[28]7.9040,[29]8.1177, save_imatrix: stored collected data after 30 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [30]8.1330,[31]8.2365,[32]8.3080,[33]8.3659,[34]8.2829,[35]8.1805,[36]7.8166,[37]7.6196,[38]7.2773,[39]7.2020, save_imatrix: stored collected data after 40 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [40]7.1117,[41]7.0423,[42]7.0131,[43]7.0513,[44]7.0433,[45]6.9766,[46]7.0585,[47]7.1493,[48]7.3117,[49]7.3431, save_imatrix: stored collected data after 50 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [50]7.5558,[51]7.7420,[52]7.9529,[53]8.1337,[54]8.2862,[55]8.2058,[56]8.2031,[57]8.3415,[58]8.4436,[59]8.4034, save_imatrix: stored collected data after 60 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [60]8.3390,[61]8.3636,[62]8.4492,[63]8.5299,[64]8.6364,[65]8.7142,[66]8.7962,[67]8.8080,[68]8.8470,[69]8.8577, save_imatrix: stored collected data after 70 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [70]8.8172,[71]8.7114,[72]8.5987,[73]8.5845,[74]8.6847,[75]8.7115,[76]8.6591,[77]8.6226,[78]8.6239,[79]8.5747, save_imatrix: stored collected data after 80 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [80]8.4730,[81]8.2890,[82]8.1267,[83]8.1267,[84]8.1288,[85]8.1302,[86]8.0806,[87]8.0983,[88]8.0851,[89]8.0748, save_imatrix: stored collected data after 90 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [90]8.0567,[91]8.0261,[92]8.0150,[93]7.9759,[94]7.8792,[95]7.9289,[96]7.9894,[97]8.0096,[98]7.9770,[99]7.9935, save_imatrix: stored collected data after 100 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [100]8.0289,[101]7.9490,[102]7.8935,[103]7.8883,[104]7.9199,[105]7.9198,[106]7.9549,[107]7.9926,[108]7.9428,[109]7.8852, save_imatrix: stored collected data after 110 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [110]7.8331,[111]7.7686,[112]7.6997,[113]7.6362,[114]7.5856,[115]7.5369,[116]7.5122,[117]7.5222,[118]7.5262,[119]7.6019, save_imatrix: stored collected data after 120 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [120]7.6779,[121]7.7421,[122]7.8017,[123]7.8961,[124]7.8799,[125]7.9118,[126]7.8995,[127]7.9356,[128]7.9227,[129]7.9105, save_imatrix: stored collected data after 130 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [130]7.8760,[131]7.8445,[132]7.8829,[133]7.9431,[134]7.9423,[135]7.9438,[136]7.9568,[137]7.9834,[138]7.9895,[139]7.9990, save_imatrix: stored collected data after 140 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat [140]8.0310,[141]8.0578,[142]8.0334,[143]8.0987,[144]8.1383,[145]8.1734,[146]8.2294,[147]8.2535,[148]8.3020,[149]8.3237, save_imatrix: stored collected data after 149 chunks in neo_7b_sft_v0.1-IMat-GGUF/imatrix.dat llama_print_timings: load time = 3161.43 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 = 124844.08 ms / 76288 tokens ( 1.64 ms per token, 611.07 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 = 127790.46 ms / 76289 tokens Final estimate: PPL = 8.3237 +/- 0.12004