Bug: infill reference crashed

[kidoln]

What happened?

./llama-infill -t 10 -ngl 0 -m ../../models/Publisher/Repository/codellama-13b.Q3_K_S.gguf --temp 0.7 --repeat_penalty 1.1 -n 20 --in-prefix "def helloworld():\n print("hell" --in-suffix "\n print("goodbye world")\n "

this command causes llama.cpp abort

Name and Version

./llama-llava-cli --version
version: 3235 (8854044)
built with Apple clang version 15.0.0 (clang-1500.3.9.4) for arm64-apple-darwin23.5.0

What operating system are you seeing the problem on?

Mac

Relevant log output

Log start
main: build = 3235 (88540445)
main: built with Apple clang version 15.0.0 (clang-1500.3.9.4) for arm64-apple-darwin23.5.0
main: seed  = 1719410022
llama_model_loader: loaded meta data with 20 key-value pairs and 363 tensors from ../../models/Publisher/Repository/codellama-13b.Q3_K_S.gguf (version GGUF V2)
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              = codellama_codellama-13b-hf
llama_model_loader: - kv   2:                       llama.context_length u32              = 16384
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 5120
llama_model_loader: - kv   4:                          llama.block_count u32              = 40
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 13824
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 40
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 40
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  11:                          general.file_type u32              = 11
llama_model_loader: - kv  12:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  13:                      tokenizer.ggml.tokens arr[str,32016]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  14:                      tokenizer.ggml.scores arr[f32,32016]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,32016]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  16:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  17:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  18:            tokenizer.ggml.unknown_token_id u32              = 0
llama_model_loader: - kv  19:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   81 tensors
llama_model_loader: - type q3_K:  281 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special tokens cache size = 259
llm_load_vocab: token to piece cache size = 0.1686 MB
llm_load_print_meta: format           = GGUF V2
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32016
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 16384
llm_load_print_meta: n_embd           = 5120
llm_load_print_meta: n_head           = 40
llm_load_print_meta: n_head_kv        = 40
llm_load_print_meta: n_layer          = 40
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 1
llm_load_print_meta: n_embd_k_gqa     = 5120
llm_load_print_meta: n_embd_v_gqa     = 5120
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             = 13824
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  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 16384
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       = 13B
llm_load_print_meta: model ftype      = Q3_K - Small
llm_load_print_meta: model params     = 13.02 B
llm_load_print_meta: model size       = 5.27 GiB (3.48 BPW)
llm_load_print_meta: general.name     = codellama_codellama-13b-hf
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
llm_load_print_meta: max token length = 48
llm_load_tensors: ggml ctx size =    0.17 MiB
llm_load_tensors: offloading 0 repeating layers to GPU
llm_load_tensors: offloaded 0/41 layers to GPU
llm_load_tensors:        CPU buffer size =  5396.21 MiB
...................................................................................................
llama_new_context_with_model: n_ctx      = 16384
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  = 1000000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init:        CPU KV buffer size = 12800.00 MiB
llama_new_context_with_model: KV self size  = 12800.00 MiB, K (f16): 6400.00 MiB, V (f16): 6400.00 MiB
llama_new_context_with_model:        CPU  output buffer size =     0.12 MiB
llama_new_context_with_model:        CPU compute buffer size =  1352.01 MiB
llama_new_context_with_model: graph nodes  = 1286
llama_new_context_with_model: graph splits = 642

system_info: n_threads = 10 / 11 | AVX = 0 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 1 | SVE = 0 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | SSSE3 = 0 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 |
libc++abi: terminating due to uncaught exception of type std::out_of_range: vector
[1]    37264 abort      ./llama-infill -t 10 -ngl 0 -m  --temp 0.7 --repeat_penalty 1.1 -n 20

[CISC]

My bad, submitted a fix, in the meantime you can fix this by adding the appropriate metadata to the GGUF:

gguf-new-metadata.py input.gguf output.gguf --special-token prefix "▁<PRE>" --special-token middle "▁<MID>" --special-token suffix "▁<SUF>" --special-token eot "▁<EOT>"

[kidoln]

My bad, submitted a fix, in the meantime you can fix this by adding the appropriate metadata to the GGUF:

gguf-new-metadata.py input.gguf output.gguf --special-token prefix "▁<PRE>" --special-token middle "▁<MID>" --special-token suffix "▁<SUF>" --special-token eot "▁<EOT>"

Thanks, But when I try to convert this model, codeshell-chat-q4_0.gguf. I received the following error.

INFO:gguf-new-metadata:* Loading: codeshell-chat-q4_0.gguf
Traceback (most recent call last):
  File "/Users/kido/Code/models/Publisher/Repository/../../../githubs/llama.cpp/gguf-py/scripts/gguf-new-metadata.py", line 242, in <module>
    main()
  File "/Users/kido/Code/models/Publisher/Repository/../../../githubs/llama.cpp/gguf-py/scripts/gguf-new-metadata.py", line 201, in main
    ids = find_token(token_list, token)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kido/Code/models/Publisher/Repository/../../../githubs/llama.cpp/gguf-py/scripts/gguf-new-metadata.py", line 73, in find_token
    raise LookupError(f'Unable to find "{token}" in token list!')
LookupError: Unable to find "▁<PRE>" in token list!

[CISC]

That's because that model has completely different FIM tokens (and no EOT token), see tokenizer_config.json, for this model you need the following:

gguf-new-metadata.py input.gguf output.gguf --special-token prefix "<fim_prefix>" --special-token middle "<fim_middle>" --special-token suffix "<fim_suffix>"

[kidoln]

./llama-infill -t 10 -ngl 0 -m ../../models/Publisher/Repository/codellama-13b.Q3_K_S.gguf --temp 0.7 --repeat_penalty 1.1 -n 20 --in-prefix "def helloworld():\n print("hell" --in-suffix "\n print("goodbye world")\n "

That fix the metadata, but I received segmentation fault during llama-infill calling.

./llama-infill -t 10 -m ../../models/Publisher/Repository/codeshell_modified.gguf --temp 0.7 --repeat_penalty 1.1 -n 20 --in-prefix "def helloworld()"
Log start
main: build = 3235 (88540445)
main: built with Apple clang version 15.0.0 (clang-1500.3.9.4) for arm64-apple-darwin23.5.0
main: seed  = 1719505502
llama_model_loader: loaded meta data with 25 key-value pairs and 508 tensors from ../../models/Publisher/Repository/codeshell_modified.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              = codeshell
llama_model_loader: - kv   1:                               general.name str              = CodeShell
llama_model_loader: - kv   2:                   codeshell.context_length u32              = 8192
llama_model_loader: - kv   3:                 codeshell.embedding_length u32              = 4096
llama_model_loader: - kv   4:              codeshell.feed_forward_length u32              = 16384
llama_model_loader: - kv   5:                      codeshell.block_count u32              = 42
llama_model_loader: - kv   6:             codeshell.attention.head_count u32              = 32
llama_model_loader: - kv   7:          codeshell.attention.head_count_kv u32              = 8
llama_model_loader: - kv   8:     codeshell.attention.layer_norm_epsilon f32              = 0.000010
llama_model_loader: - kv   9:                          general.file_type u32              = 2
llama_model_loader: - kv  10:                   codeshell.rope.freq_base f32              = 10000.000000
llama_model_loader: - kv  11:                codeshell.rope.scale_linear f32              = 1.000000
llama_model_loader: - kv  12:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  13:                      tokenizer.ggml.tokens arr[str,70144]   = ["æ½»", "æ¶ģ", "ïĴĻ", "amily...
llama_model_loader: - kv  14:                      tokenizer.ggml.scores arr[f32,70144]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,70144]   = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  16:                      tokenizer.ggml.merges arr[str,72075]   = ["Ġ Ġ", "ĠĠ ĠĠ", "ĠĠĠĠ ĠĠ...
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 70000
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 70000
llama_model_loader: - kv  19:            tokenizer.ggml.unknown_token_id u32              = 70000
llama_model_loader: - kv  20:            tokenizer.ggml.padding_token_id u32              = 70000
llama_model_loader: - kv  21:               general.quantization_version u32              = 2
llama_model_loader: - kv  22:             tokenizer.ggml.prefix_token_id u32              = 70001
llama_model_loader: - kv  23:             tokenizer.ggml.middle_token_id u32              = 70002
llama_model_loader: - kv  24:             tokenizer.ggml.suffix_token_id u32              = 70003
llama_model_loader: - type  f32:  338 tensors
llama_model_loader: - type q4_0:  169 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: missing pre-tokenizer type, using: 'default'
llm_load_vocab:
llm_load_vocab: ************************************
llm_load_vocab: GENERATION QUALITY WILL BE DEGRADED!
llm_load_vocab: CONSIDER REGENERATING THE MODEL
llm_load_vocab: ************************************
llm_load_vocab:
llm_load_vocab: special tokens cache size = 0
llm_load_vocab: token to piece cache size = 0.2985 MB
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = codeshell
llm_load_print_meta: vocab type       = BPE
llm_load_print_meta: n_vocab          = 70144
llm_load_print_meta: n_merges         = 72075
llm_load_print_meta: n_ctx_train      = 8192
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 42
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 1.0e-05
llm_load_print_meta: f_norm_rms_eps   = 0.0e+00
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             = 16384
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_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       = 0.1B
llm_load_print_meta: model ftype      = Q4_0
llm_load_print_meta: model params     = 7.98 B
llm_load_print_meta: model size       = 4.25 GiB (4.58 BPW)
llm_load_print_meta: general.name     = CodeShell
llm_load_print_meta: BOS token        = 70000 '<|endoftext|>'
llm_load_print_meta: EOS token        = 70000 '<|endoftext|>'
llm_load_print_meta: UNK token        = 70000 '<|endoftext|>'
llm_load_print_meta: PAD token        = 70000 '<|endoftext|>'
llm_load_print_meta: LF token         = 28544 'ÄĬ'
llm_load_print_meta: PRE token        = 70001 '<fim_prefix>'
llm_load_print_meta: SUF token        = 70003 '<fim_suffix>'
llm_load_print_meta: MID token        = 70002 '<fim_middle>'
llm_load_print_meta: EOT token        = 70000 '<|endoftext|>'
llm_load_print_meta: max token length = 256
llm_load_tensors: ggml ctx size =    0.45 MiB
ggml_backend_metal_log_allocated_size: allocated buffer, size =  4201.36 MiB, ( 4201.44 / 12288.02)
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:      Metal buffer size =  4201.35 MiB
llm_load_tensors:        CPU buffer size =   154.12 MiB
.............................................................................................
llama_new_context_with_model: n_ctx      = 8192
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
ggml_metal_init: allocating
ggml_metal_init: found device: Apple M3 Pro
ggml_metal_init: picking default device: Apple M3 Pro
ggml_metal_init: default.metallib not found, loading from source
ggml_metal_init: GGML_METAL_PATH_RESOURCES = nil
ggml_metal_init: loading '/Users/kido/Code/githubs/llama.cpp/ggml-metal.metal'
ggml_metal_init: GPU name:   Apple M3 Pro
ggml_metal_init: GPU family: MTLGPUFamilyApple9  (1009)
ggml_metal_init: GPU family: MTLGPUFamilyCommon3 (3003)
ggml_metal_init: GPU family: MTLGPUFamilyMetal3  (5001)
ggml_metal_init: simdgroup reduction support   = true
ggml_metal_init: simdgroup matrix mul. support = true
ggml_metal_init: hasUnifiedMemory              = true
ggml_metal_init: recommendedMaxWorkingSetSize  = 12884.92 MB
llama_kv_cache_init:      Metal KV buffer size =  1344.00 MiB
llama_new_context_with_model: KV self size  = 1344.00 MiB, K (f16):  672.00 MiB, V (f16):  672.00 MiB
llama_new_context_with_model:        CPU  output buffer size =     0.27 MiB
llama_new_context_with_model:      Metal compute buffer size =   564.00 MiB
llama_new_context_with_model:        CPU compute buffer size =    24.01 MiB
llama_new_context_with_model: graph nodes  = 1687
llama_new_context_with_model: graph splits = 2

system_info: n_threads = 10 / 11 | AVX = 0 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 1 | SVE = 0 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 1 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | SSSE3 = 0 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 |
[1]    32542 segmentation fault  ./llama-infill -t 10 -m  --temp 0.7 --repeat_penalty 1.1 -n 20 --in-prefix

[CISC]

The infill example is not very stable, it's missing a few checks, my guess is it's because you're missing --in-suffix. It will also crash on models with no EOT, but only after outputting the result.

Please submit another issue.

[kidoln]

--in-prefix "def helloworld():\n print("hell" --in-suffix "\n print("goodbye world")\n "

yes, adding --in-suffix fix the problem