Eval bug: process crashes when using the LFM2-VL-3B model for Regenerate #17043
Description
Name and Version
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 GTX 1050 Ti with Max-Q Design, compute capability 6.1, VMM: yes
load_backend: loaded CUDA backend from C:\gguf-models\llama.cpp\ggml-cuda.dll
load_backend: loaded RPC backend from C:\gguf-models\llama.cpp\ggml-rpc.dll
load_backend: loaded CPU backend from C:\gguf-models\llama.cpp\ggml-cpu-haswell.dll
version: 6949 (a5c07dcd7)
built with clang version 19.1.5 for x86_64-pc-windows-msvc
Operating systems
Windows
GGML backends
CUDA
Hardware
NVIDIA GeForce GTX 1050 Ti
Models
Problem description & steps to reproduce
When using the llama-server to load the LFM2-VL-3B model for inference, clicking the "Regenerate" button causes the server process to crash. Printing D:/a/llama.cpp/llama.cpp/tools/server/server.cpp:3885: GGML_ASSERT(!slot.prompt.tokens.has_mtmd) failed reveals the issue.
I have tried various parameters such as -t 1 --no-mmap --mlock --cache-ram 0 --no-mmproj-offload, but they have not prevented this problem.
First Bad Commit
No response
Relevant log output
llama-server -m ../models/LFM2-VL-3B.Q8_0.gguf --mmproj ../models/LFM2-VL-3B.mmproj.f16.gguf
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 GTX 1050 Ti with Max-Q Design, compute capability 6.1, VMM: yes
load_backend: loaded CUDA backend from C:\gguf-models\llama.cpp\ggml-cuda.dll
load_backend: loaded RPC backend from C:\gguf-models\llama.cpp\ggml-rpc.dll
load_backend: loaded CPU backend from C:\gguf-models\llama.cpp\ggml-cpu-haswell.dll
main: setting n_parallel = 4 and kv_unified = true
build: 6949 (a5c07dcd7) with clang version 19.1.5 for x86_64-pc-windows-msvc
system info: n_threads = 6, n_threads_batch = 6, total_threads = 6
system_info: n_threads = 6 (n_threads_batch = 6) / 6 | CUDA : ARCHS = 500,610,700,750,800,860,890 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA
= 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 |
main: binding port with default address family
main: HTTP server is listening, hostname: 127.0.0.1, port: 8080, http threads: 6
main: loading model
srv load_model: loading model '../models/LFM2-VL-3B.Q8_0.gguf'
llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce GTX 1050 Ti with Max-Q Design) (0000:01:00.0) - 3359 MiB free
llama_model_loader: loaded meta data with 34 key-value pairs and 266 tensors from ../models/LFM2-VL-3B.Q8_0.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 = lfm2
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.name str = LFM2 VL 3B
llama_model_loader: - kv 3: general.basename str = LFM2-VL
llama_model_loader: - kv 4: general.size_label str = 3B
llama_model_loader: - kv 5: general.license str = other
llama_model_loader: - kv 6: general.license.name str = lfm1.0
llama_model_loader: - kv 7: general.license.link str = LICENSE
llama_model_loader: - kv 8: general.tags arr[str,5] = ["liquid", "lfm2", "lfm2-vl", "edge",...
llama_model_loader: - kv 9: general.languages arr[str,1] = ["en"]
llama_model_loader: - kv 10: lfm2.block_count u32 = 30
llama_model_loader: - kv 11: lfm2.context_length u32 = 128000
llama_model_loader: - kv 12: lfm2.embedding_length u32 = 2048
llama_model_loader: - kv 13: lfm2.feed_forward_length u32 = 10752
llama_model_loader: - kv 14: lfm2.attention.head_count u32 = 32
llama_model_loader: - kv 15: lfm2.attention.head_count_kv arr[i32,30] = [0, 0, 8, 0, 0, 8, 0, 0, 0, 8, 0, 0, ...
llama_model_loader: - kv 16: lfm2.rope.freq_base f32 = 1000000.000000
llama_model_loader: - kv 17: lfm2.attention.layer_norm_rms_epsilon f32 = 0.000010
llama_model_loader: - kv 18: lfm2.vocab_size u32 = 65536
llama_model_loader: - kv 19: lfm2.shortconv.l_cache u32 = 3
llama_model_loader: - kv 20: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 21: tokenizer.ggml.pre str = lfm2
llama_model_loader: - kv 22: tokenizer.ggml.tokens arr[str,65536] = ["<|pad|>", "<|startoftext|>", "<|end...
llama_model_loader: - kv 23: tokenizer.ggml.token_type arr[i32,65536] = [3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, ...
llama_model_loader: - kv 24: tokenizer.ggml.merges arr[str,63683] = ["Ċ Ċ", "Ċ ĊĊ", "ĊĊ Ċ", "Ċ ...
llama_model_loader: - kv 25: tokenizer.ggml.bos_token_id u32 = 1
llama_model_loader: - kv 26: tokenizer.ggml.eos_token_id u32 = 7
llama_model_loader: - kv 27: tokenizer.ggml.padding_token_id u32 = 0
llama_model_loader: - kv 28: tokenizer.ggml.add_bos_token bool = true
llama_model_loader: - kv 29: tokenizer.ggml.add_sep_token bool = false
llama_model_loader: - kv 30: tokenizer.ggml.add_eos_token bool = false
llama_model_loader: - kv 31: tokenizer.chat_template str = {{bos_token}}{% for message in messag...
llama_model_loader: - kv 32: general.quantization_version u32 = 2
llama_model_loader: - kv 33: general.file_type u32 = 7
llama_model_loader: - type f32: 99 tensors
llama_model_loader: - type q8_0: 167 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q8_0
print_info: file size = 2.54 GiB (8.50 BPW)
load: printing all EOG tokens:
load: - 2 ('<|endoftext|>')
load: - 7 ('<|im_end|>')
load: special tokens cache size = 507
load: token to piece cache size = 0.3759 MB
print_info: arch = lfm2
print_info: vocab_only = 0
print_info: n_ctx_train = 128000
print_info: n_embd = 2048
print_info: n_layer = 30
print_info: n_head = 32
print_info: n_head_kv = [0, 0, 8, 0, 0, 8, 0, 0, 0, 8, 0, 0, 0, 8, 0, 0, 0, 8, 0, 0, 0, 8, 0, 0, 8, 0, 0, 8, 0, 0]
print_info: n_rot = 64
print_info: n_swa = 0
print_info: is_swa_any = 0
print_info: n_embd_head_k = 64
print_info: n_embd_head_v = 64
print_info: n_gqa = [0, 0, 4, 0, 0, 4, 0, 0, 0, 4, 0, 0, 0, 4, 0, 0, 0, 4, 0, 0, 0, 4, 0, 0, 4, 0, 0, 4, 0, 0]
print_info: n_embd_k_gqa = [0, 0, 512, 0, 0, 512, 0, 0, 0, 512, 0, 0, 0, 512, 0, 0, 0, 512, 0, 0, 0, 512, 0, 0, 512, 0, 0, 512, 0, 0]
print_info: n_embd_v_gqa = [0, 0, 512, 0, 0, 512, 0, 0, 0, 512, 0, 0, 0, 512, 0, 0, 0, 512, 0, 0, 0, 512, 0, 0, 512, 0, 0, 512, 0, 0]
print_info: f_norm_eps = 0.0e+00
print_info: f_norm_rms_eps = 1.0e-05
print_info: f_clamp_kqv = 0.0e+00
print_info: f_max_alibi_bias = 0.0e+00
print_info: f_logit_scale = 0.0e+00
print_info: f_attn_scale = 0.0e+00
print_info: n_ff = 10752
print_info: n_expert = 0
print_info: n_expert_used = 0
print_info: n_expert_groups = 0
print_info: n_group_used = 0
print_info: causal attn = 1
print_info: pooling type = 0
print_info: rope type = 2
print_info: rope scaling = linear
print_info: freq_base_train = 1000000.0
print_info: freq_scale_train = 1
print_info: n_ctx_orig_yarn = 128000
print_info: rope_finetuned = unknown
print_info: model type = 2.6B
print_info: model params = 2.57 B
print_info: general.name = LFM2 VL 3B
print_info: vocab type = BPE
print_info: n_vocab = 65536
print_info: n_merges = 63683
print_info: BOS token = 1 '<|startoftext|>'
print_info: EOS token = 7 '<|im_end|>'
print_info: EOT token = 7 '<|im_end|>'
print_info: PAD token = 0 '<|pad|>'
print_info: LF token = 708 'Ċ'
print_info: EOG token = 2 '<|endoftext|>'
print_info: EOG token = 7 '<|im_end|>'
print_info: max token length = 30
load_tensors: loading model tensors, this can take a while... (mmap = true)
load_tensors: offloading 30 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 31/31 layers to GPU
load_tensors: CPU_Mapped model buffer size = 136.01 MiB
load_tensors: CUDA0 model buffer size = 2604.11 MiB
.............................................................................................
llama_context: constructing llama_context
llama_context: n_seq_max = 4
llama_context: n_ctx = 4096
llama_context: n_ctx_seq = 4096
llama_context: n_batch = 2048
llama_context: n_ubatch = 512
llama_context: causal_attn = 1
llama_context: flash_attn = auto
llama_context: kv_unified = true
llama_context: freq_base = 1000000.0
llama_context: freq_scale = 1
llama_context: n_ctx_seq (4096) < n_ctx_train (128000) -- the full capacity of the model will not be utilized
llama_context: CUDA_Host output buffer size = 1.00 MiB
llama_kv_cache: CUDA0 KV buffer size = 64.00 MiB
llama_kv_cache: size = 64.00 MiB ( 4096 cells, 8 layers, 4/1 seqs), K (f16): 32.00 MiB, V (f16): 32.00 MiB
llama_memory_recurrent: CUDA0 RS buffer size = 1.38 MiB
llama_memory_recurrent: size = 1.38 MiB ( 4 cells, 30 layers, 4 seqs), R (f32): 1.38 MiB, S (f32): 0.00 MiB
llama_context: Flash Attention was auto, set to enabled
llama_context: CUDA0 compute buffer size = 136.02 MiB
llama_context: CUDA_Host compute buffer size = 12.01 MiB
llama_context: graph nodes = 1015
llama_context: graph splits = 4 (with bs=512), 5 (with bs=1)
common_init_from_params: added <|endoftext|> logit bias = -inf
common_init_from_params: added <|im_end|> logit bias = -inf
common_init_from_params: setting dry_penalty_last_n to ctx_size = 4096
common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
clip_model_loader: model name: LFM2 VL 3B
clip_model_loader: description:
clip_model_loader: GGUF version: 3
clip_model_loader: alignment: 32
clip_model_loader: n_tensors: 443
clip_model_loader: n_kv: 27
clip_model_loader: has vision encoder
clip_ctx: CLIP using CUDA0 backend
load_hparams: projector: lfm2
load_hparams: n_embd: 1152
load_hparams: n_head: 16
load_hparams: n_ff: 4304
load_hparams: n_layer: 27
load_hparams: ffn_op: gelu
load_hparams: projection_dim: 2048
--- vision hparams ---
load_hparams: image_size: 256
load_hparams: patch_size: 16
load_hparams: has_llava_proj: 0
load_hparams: minicpmv_version: 0
load_hparams: n_merge: 2
load_hparams: n_wa_pattern: 0
load_hparams: image_min_pixels: 65536
load_hparams: image_max_pixels: 262144
load_hparams: model size: 820.94 MiB
load_hparams: metadata size: 0.16 MiB
alloc_compute_meta: warmup with image size = 512 x 512
alloc_compute_meta: CUDA0 compute buffer size = 30.31 MiB
alloc_compute_meta: CPU compute buffer size = 3.00 MiB
alloc_compute_meta: graph splits = 1, nodes = 871
warmup: flash attention is enabled
srv load_model: loaded multimodal model, '../models/LFM2-VL-3B.mmproj.f16.gguf'
srv init: initializing slots, n_slots = 4
slot init: id 0 | task -1 | new slot, n_ctx = 4096
slot init: id 1 | task -1 | new slot, n_ctx = 4096
slot init: id 2 | task -1 | new slot, n_ctx = 4096
slot init: id 3 | task -1 | new slot, n_ctx = 4096
srv init: prompt cache is enabled, size limit: 8192 MiB
srv init: use `--cache-ram 0` to disable the prompt cache
srv init: for more info see https://github.com/ggml-org/llama.cpp/pull/16391
srv init: thinking = 0
main: model loaded
main: chat template, chat_template: {{bos_token}}{% for message in messages %}{{'<|im_start|>' + message['role'] + '
'}}{% if message['content'] is string %}{{ message['content'] }}{% else %}{% for content in message['content'] %}{% if content['type'] == 'image' %}{{ '<image>' }}{% elif content['type'] == 'text' %}{{ content[
'text'] }}{% endif %}{% endfor %}{% endif %}{{'<|im_end|>
'}}{% endfor %}{% if add_generation_prompt %}{{'<|im_start|>assistant
' }}{% endif %}, example_format: '<|im_start|>system
You are a helpful assistant<|im_end|>
<|im_start|>user
Hello<|im_end|>
<|im_start|>assistant
Hi there<|im_end|>
<|im_start|>user
How are you?<|im_end|>
<|im_start|>assistant
'
main: server is listening on http://127.0.0.1:8080 - starting the main loop
srv update_slots: all slots are idle
srv params_from_: Chat format: Content-only
slot get_availabl: id 3 | task -1 | selected slot by LRU, t_last = -1
slot launch_slot_: id 3 | task 0 | processing task
slot update_slots: id 3 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 0, task.n_tokens = 11
slot update_slots: id 3 | task 0 | n_tokens = 0, memory_seq_rm [0, end)
slot update_slots: id 3 | task 0 | prompt processing progress, n_tokens = 11, batch.n_tokens = 11, progress = 1.000000
slot update_slots: id 3 | task 0 | prompt done, n_tokens = 11, batch.n_tokens = 11
srv log_server_r: request: GET /props 127.0.0.1 200
srv log_server_r: request: GET /slots 127.0.0.1 200
srv log_server_r: request: GET /slots 127.0.0.1 200
srv log_server_r: request: GET /slots 127.0.0.1 200
slot print_timing: id 3 | task 0 |
prompt eval time = 99.13 ms / 11 tokens ( 9.01 ms per token, 110.96 tokens per second)
eval time = 395.72 ms / 10 tokens ( 39.57 ms per token, 25.27 tokens per second)
total time = 494.85 ms / 21 tokens
slot release: id 3 | task 0 | stop processing: n_tokens = 20, truncated = 0
srv update_slots: all slots are idle
srv log_server_r: request: POST /v1/chat/completions 127.0.0.1 200
srv update_slots: all slots are idle
srv log_server_r: request: GET /slots 127.0.0.1 200
srv params_from_: Chat format: Content-only
slot get_availabl: id 3 | task -1 | selected slot by LCP similarity, sim_best = 1.000 (> 0.100 thold), f_keep = 0.550
slot launch_slot_: id 3 | task 15 | processing task
slot update_slots: id 3 | task 15 | new prompt, n_ctx_slot = 4096, n_keep = 0, task.n_tokens = 11
D:/a/llama.cpp/llama.cpp/tools/server/server.cpp:3885: GGML_ASSERT(!slot.prompt.tokens.has_mtmd) failed