Bug: gemma2 perplexity pending forever

What happened?

I am attempting to measure the perplexity of the gemma-2-9b-it-Q4_K_M.gguf model using llama.cpp. However, I encounter an issue where the process gets stuck at the "tokenizing the input" stage indefinitely.

I have confirmed that the qwen2-7b-instruct-q4_k_m.gguf model operates correctly in the same environment, so I expected gemma-2 to function properly as well. Unfortunately, it does not.

the model is from huggingface model hub, bartowski/gemma-2-9b-it-GGUF

more information

I just found out that the data I have is a Korean Wikipedia dataset, and it worked fine with qwen2, but it doesn't seem to work with gemma2. After changing the data to a wiki.test.raw file, I confirmed that it works properly

I also discovered that the original number of files was 10,000, but after reducing it to 500, it worked. It seems to operate much slower compared to Qwen.

지미 카터 Introduction

'''제임스 얼 “지미” 카터 주니어'''(, 1924년 10월 1일~)는 민주당 출신 미국의 제39대 대통령 (1977-81)이다. 약력

1. 조지아 주 의회 상원의원
1971.1. 제76대 조지아 주지사 (1971.1.12.~1975.1.14.)
1977.1. 제39대 미국 대통령 (1977.1.20.~1981.1.20.)
1. 미국 대통령 자유 훈장 수훈
1. 노벨 평화상 수상
1. 제49회 그래미 어워드 최고의 낭독 앨범상 수상
1. 제58회 그래미 어워드 최고의 낭독 앨범상 수상 생애 === 어린 시절 === 지미 카터는 조지아주 섬터 카운티 플레인스 마을에서 태어났다.

조지아 공과대학교를 졸업하였다. 그 후 해군에 들어가 전함·원자력·잠수함의 승무원으로 일하였다. 1953년 미국 해군 대위로 예편하였고 이후 땅콩·면화 등을 가꿔 많은 돈을 벌었다. 그의 별명이 "땅콩 농부" (Peanut Farmer)로 알려졌다.

Name and Version

version: 3392 (bda62d79) built with cc (Ubuntu 11.3.0-1ubuntu1~22.04.1) 11.3.0 for x86_64-linux-gnu

What operating system are you seeing the problem on?

Linux

Relevant log output

./llama-perplexity -m /home/jovyan/python-project/malpyeong2024/models/gemma2/gemma-2-9b-it-Q4_K_M.gguf -f /home/jovyan/python-project/malpyeong2024/korean-wikipedia-train.txt -ngl 100

main: build = 3392 (bda62d79)
main: built with cc (Ubuntu 11.3.0-1ubuntu1~22.04.1) 11.3.0 for x86_64-linux-gnu
main: seed  = 1721041506
llama_model_loader: loaded meta data with 33 key-value pairs and 464 tensors from /home/jovyan/python-project/malpyeong2024/models/gemma2/gemma-2-9b-it-Q4_K_M.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.name str              = gemma-2-9b-it
llama_model_loader: - kv   2:                      gemma2.context_length u32              = 8192
llama_model_loader: - kv   3:                    gemma2.embedding_length u32              = 3584
llama_model_loader: - kv   4:                         gemma2.block_count u32              = 42
llama_model_loader: - kv   5:                 gemma2.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                gemma2.attention.head_count u32              = 16
llama_model_loader: - kv   7:             gemma2.attention.head_count_kv u32              = 8
llama_model_loader: - kv   8:    gemma2.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv   9:                gemma2.attention.key_length u32              = 256
llama_model_loader: - kv  10:              gemma2.attention.value_length u32              = 256
llama_model_loader: - kv  11:                          general.file_type u32              = 15
llama_model_loader: - kv  12:              gemma2.attn_logit_softcapping f32              = 50.000000
llama_model_loader: - kv  13:             gemma2.final_logit_softcapping f32              = 30.000000
llama_model_loader: - kv  14:            gemma2.attention.sliding_window u32              = 4096
llama_model_loader: - kv  15:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  16:                         tokenizer.ggml.pre str              = default
llama_model_loader: - kv  17:                      tokenizer.ggml.tokens arr[str,256000]  = ["<pad>", "<eos>", "<bos>", "<unk>", ...
llama_model_loader: - kv  18:                      tokenizer.ggml.scores arr[f32,256000]  = [-1000.000000, -1000.000000, -1000.00...
llama_model_loader: - kv  19:                  tokenizer.ggml.token_type arr[i32,256000]  = [3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, ...
llama_model_loader: - kv  20:                tokenizer.ggml.bos_token_id u32              = 2
llama_model_loader: - kv  21:                tokenizer.ggml.eos_token_id u32              = 1
llama_model_loader: - kv  22:            tokenizer.ggml.unknown_token_id u32              = 3
llama_model_loader: - kv  23:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  24:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  25:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  26:                    tokenizer.chat_template str              = {{ bos_token }}{% if messages[0]['rol...
llama_model_loader: - kv  27:            tokenizer.ggml.add_space_prefix bool             = false
llama_model_loader: - kv  28:               general.quantization_version u32              = 2
llama_model_loader: - kv  29:                      quantize.imatrix.file str              = /models_out/gemma-2-9b-it-GGUF/gemma-...
llama_model_loader: - kv  30:                   quantize.imatrix.dataset str              = /training_dir/calibration_datav3.txt
llama_model_loader: - kv  31:             quantize.imatrix.entries_count i32              = 294
llama_model_loader: - kv  32:              quantize.imatrix.chunks_count i32              = 128
llama_model_loader: - type  f32:  169 tensors
llama_model_loader: - type q4_K:  252 tensors
llama_model_loader: - type q6_K:   43 tensors
llm_load_vocab: special tokens cache size = 217
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           = 3584
llm_load_print_meta: n_layer          = 42
llm_load_print_meta: n_head           = 16
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_rot            = 256
llm_load_print_meta: n_swa            = 4096
llm_load_print_meta: n_embd_head_k    = 256
llm_load_print_meta: n_embd_head_v    = 256
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             = 14336
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       = 9B
llm_load_print_meta: model ftype      = Q4_K - Medium
llm_load_print_meta: model params     = 9.24 B
llm_load_print_meta: model size       = 5.36 GiB (4.98 BPW) 
llm_load_print_meta: general.name     = gemma-2-9b-it
llm_load_print_meta: BOS token        = 2 '<bos>'
llm_load_print_meta: EOS token        = 1 '<eos>'
llm_load_print_meta: UNK token        = 3 '<unk>'
llm_load_print_meta: PAD token        = 0 '<pad>'
llm_load_print_meta: LF token         = 227 '<0x0A>'
llm_load_print_meta: EOT token        = 107 '<end_of_turn>'
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 2 CUDA devices:
  Device 0: NVIDIA RTX A6000, compute capability 8.6, VMM: yes
  Device 1: NVIDIA RTX A6000, compute capability 8.6, VMM: yes
llm_load_tensors: ggml ctx size =    0.62 MiB
llm_load_tensors: offloading 42 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 43/43 layers to GPU
llm_load_tensors:        CPU buffer size =   717.77 MiB
llm_load_tensors:      CUDA0 buffer size =  2484.45 MiB
llm_load_tensors:      CUDA1 buffer size =  3003.94 MiB
...............................................................................
llama_new_context_with_model: n_ctx      = 2048
llama_new_context_with_model: n_batch    = 2048
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 =   352.00 MiB
llama_kv_cache_init:      CUDA1 KV buffer size =   320.00 MiB
llama_new_context_with_model: KV self size  =  672.00 MiB, K (f16):  336.00 MiB, V (f16):  336.00 MiB
llama_new_context_with_model:  CUDA_Host  output buffer size =     3.91 MiB
llama_new_context_with_model: pipeline parallelism enabled (n_copies=4)
llama_new_context_with_model:      CUDA0 compute buffer size =   154.01 MiB
llama_new_context_with_model:      CUDA1 compute buffer size =   567.02 MiB
llama_new_context_with_model:  CUDA_Host compute buffer size =    39.02 MiB
llama_new_context_with_model: graph nodes  = 1690
llama_new_context_with_model: graph splits = 3

system_info: n_threads = 16 / 32 | AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | 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 | 
perplexity: tokenizing the input ..

ggerganov / llama.cpp