server: Describing pictures with multi models seems to crash the model

[domasofan]

Hi all,

Tried to describe a picture with these two models in separate runs:
https://huggingface.co/bartowski/mistralai_Mistral-Small-3.1-24B-Instruct-2503-GGUF
https://huggingface.co/bartowski/Qwen_Qwen2.5-VL-7B-Instruct-GGUF

The llama.cpp build used was b5351 CPU X64 on Win 11.
No errors where thrown.

Greetings,
Simon

[zandera007]

I have the same issue whit Rolmocr model that is a fine-tuned Qwen2.5 model. Only text prompt is fonctionnning but when I send an image with text there is a crash...

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 GeForce RTX 3060, compute capability 8.6, VMM: yes
Device 1: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes
load_backend: loaded CUDA backend from /app/libggml-cuda.so
load_backend: loaded CPU backend from /app/libggml-cpu-haswell.so
build: 5350 (c104023) with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
system info: n_threads = 4, n_threads_batch = 4, total_threads = 8
system_info: n_threads = 4 (n_threads_batch = 4) / 8 | 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 | AARCH64_REPACK = 1 |
main: binding port with default address family
main: HTTP server is listening, hostname: 0.0.0.0, port: 8080, http threads: 7
main: loading model
srv load_model: loading model '/models/RolmOCR.Q8_0.gguf'
llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce RTX 3060) - 11796 MiB free
llama_model_load_from_file_impl: using device CUDA1 (NVIDIA GeForce RTX 3060) - 11796 MiB free
llama_model_loader: loaded meta data with 42 key-value pairs and 339 tensors from /models/RolmOCR.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 = qwen2vl
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.name str = RolmOCR
llama_model_loader: - kv 3: general.size_label str = 7.6B
llama_model_loader: - kv 4: general.license str = apache-2.0
llama_model_loader: - kv 5: general.base_model.count u32 = 1
llama_model_loader: - kv 6: general.base_model.0.name str = Qwen2.5 VL 7B Instruct
llama_model_loader: - kv 7: general.base_model.0.organization str = Qwen
llama_model_loader: - kv 8: general.base_model.0.repo_url str = https://huggingface.co/Qwen/Qwen2.5-V...
llama_model_loader: - kv 9: general.dataset.count u32 = 1
llama_model_loader: - kv 10: general.dataset.0.name str = olmOCR Mix 0225
llama_model_loader: - kv 11: general.dataset.0.version str = 0225
llama_model_loader: - kv 12: general.dataset.0.organization str = Allenai
llama_model_loader: - kv 13: general.dataset.0.repo_url str = https://huggingface.co/allenai/olmOCR...
llama_model_loader: - kv 14: qwen2vl.block_count u32 = 28
llama_model_loader: - kv 15: qwen2vl.context_length u32 = 128000
llama_model_loader: - kv 16: qwen2vl.embedding_length u32 = 3584
llama_model_loader: - kv 17: qwen2vl.feed_forward_length u32 = 18944
llama_model_loader: - kv 18: qwen2vl.attention.head_count u32 = 28
llama_model_loader: - kv 19: qwen2vl.attention.head_count_kv u32 = 4
llama_model_loader: - kv 20: qwen2vl.rope.freq_base f32 = 1000000.000000
llama_model_loader: - kv 21: qwen2vl.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 22: qwen2vl.rope.dimension_sections arr[i32,4] = [16, 24, 24, 0]
llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 24: tokenizer.ggml.pre str = qwen2
llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,152064] = ["!", """, "#", "$", "%", "&", "'", ...
llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,152064] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv 28: tokenizer.ggml.eos_token_id u32 = 151645
llama_model_loader: - kv 29: tokenizer.ggml.padding_token_id u32 = 151643
llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 151643
llama_model_loader: - kv 31: tokenizer.ggml.add_bos_token bool = false
llama_model_loader: - kv 32: tokenizer.chat_template str = {%- if tools %}\n {{- '<|im_start|>...
llama_model_loader: - kv 33: general.quantization_version u32 = 2
llama_model_loader: - kv 34: general.file_type u32 = 7
llama_model_loader: - kv 35: general.url str = https://huggingface.co/mradermacher/R...
llama_model_loader: - kv 36: mradermacher.quantize_version str = 2
llama_model_loader: - kv 37: mradermacher.quantized_by str = mradermacher
llama_model_loader: - kv 38: mradermacher.quantized_at str = 2025-04-12T22:50:21+02:00
llama_model_loader: - kv 39: mradermacher.quantized_on str = nico1
llama_model_loader: - kv 40: general.source.url str = https://huggingface.co/reducto/RolmOCR
llama_model_loader: - kv 41: mradermacher.convert_type str = hf
llama_model_loader: - type f32: 141 tensors
llama_model_loader: - type q8_0: 198 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q8_0
print_info: file size = 7.54 GiB (8.50 BPW)
load: special tokens cache size = 22
load: token to piece cache size = 0.9310 MB
print_info: arch = qwen2vl
print_info: vocab_only = 0
print_info: n_ctx_train = 128000
print_info: n_embd = 3584
print_info: n_layer = 28
print_info: n_head = 28
print_info: n_head_kv = 4
print_info: n_rot = 128
print_info: n_swa = 0
print_info: n_swa_pattern = 1
print_info: n_embd_head_k = 128
print_info: n_embd_head_v = 128
print_info: n_gqa = 7
print_info: n_embd_k_gqa = 512
print_info: n_embd_v_gqa = 512
print_info: f_norm_eps = 0.0e+00
print_info: f_norm_rms_eps = 1.0e-06
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 = 18944
print_info: n_expert = 0
print_info: n_expert_used = 0
print_info: causal attn = 1
print_info: pooling type = -1
print_info: rope type = 8
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: ssm_d_conv = 0
print_info: ssm_d_inner = 0
print_info: ssm_d_state = 0
print_info: ssm_dt_rank = 0
print_info: ssm_dt_b_c_rms = 0
print_info: model type = 7B
print_info: model params = 7.62 B
print_info: general.name = RolmOCR
print_info: vocab type = BPE
print_info: n_vocab = 152064
print_info: n_merges = 151387
print_info: BOS token = 151643 '<|endoftext|>'
print_info: EOS token = 151645 '<|im_end|>'
print_info: EOT token = 151645 '<|im_end|>'
print_info: PAD token = 151643 '<|endoftext|>'
print_info: LF token = 198 'Ċ'
print_info: FIM PRE token = 151659 '<|fim_prefix|>'
print_info: FIM SUF token = 151661 '<|fim_suffix|>'
print_info: FIM MID token = 151660 '<|fim_middle|>'
print_info: FIM PAD token = 151662 '<|fim_pad|>'
print_info: FIM REP token = 151663 '<|repo_name|>'
print_info: FIM SEP token = 151664 '<|file_sep|>'
print_info: EOG token = 151643 '<|endoftext|>'
print_info: EOG token = 151645 '<|im_end|>'
print_info: EOG token = 151662 '<|fim_pad|>'
print_info: EOG token = 151663 '<|repo_name|>'
print_info: EOG token = 151664 '<|file_sep|>'
print_info: max token length = 256
load_tensors: loading model tensors, this can take a while... (mmap = true)
load_tensors: offloading 28 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 29/29 layers to GPU
load_tensors: CUDA0 model buffer size = 3542.78 MiB
load_tensors: CUDA1 model buffer size = 3622.66 MiB
load_tensors: CPU_Mapped model buffer size = 552.23 MiB
..............................................srv log_server_r: request: GET /health 127.0.0.1 503
........................................
llama_context: constructing llama_context
llama_context: n_seq_max = 1
llama_context: n_ctx = 4096
llama_context: n_ctx_per_seq = 4096
llama_context: n_batch = 2048
llama_context: n_ubatch = 512
llama_context: causal_attn = 1
llama_context: flash_attn = 0
llama_context: freq_base = 1000000.0
llama_context: freq_scale = 1
llama_context: n_ctx_per_seq (4096) < n_ctx_train (128000) -- the full capacity of the model will not be utilized
llama_context: CUDA_Host output buffer size = 0.58 MiB
llama_kv_cache_unified: kv_size = 4096, type_k = 'f16', type_v = 'f16', n_layer = 28, can_shift = 1, padding = 32
llama_kv_cache_unified: CUDA0 KV buffer size = 120.00 MiB
llama_kv_cache_unified: CUDA1 KV buffer size = 104.00 MiB
llama_kv_cache_unified: KV self size = 224.00 MiB, K (f16): 112.00 MiB, V (f16): 112.00 MiB
llama_context: pipeline parallelism enabled (n_copies=4)
llama_context: CUDA0 compute buffer size = 312.03 MiB
llama_context: CUDA1 compute buffer size = 364.04 MiB
llama_context: CUDA_Host compute buffer size = 39.05 MiB
llama_context: graph nodes = 1042
llama_context: graph splits = 3
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_ctx: CLIP using CUDA0 backend
clip_model_loader: model name: Qwen2.5-VL-7B-Instruct
clip_model_loader: description: image encoder for Qwen2VL
clip_model_loader: GGUF version: 3
clip_model_loader: alignment: 32
clip_model_loader: n_tensors: 520
clip_model_loader: n_kv: 21
load_hparams: projector: qwen2.5vl_merger
load_hparams: n_embd: 1280
load_hparams: n_head: 16
load_hparams: n_ff: 0
load_hparams: n_layer: 32
load_hparams: projection_dim: 3584
load_hparams: image_size: 3584
load_hparams: patch_size: 14
load_hparams: has_llava_proj: 0
load_hparams: minicpmv_version: 0
load_hparams: proj_scale_factor: 0
load_hparams: n_wa_pattern: 8
load_hparams: ffn_op: silu
load_hparams: model size: 1291.40 MiB
load_hparams: metadata size: 0.18 MiB
alloc_compute_meta: CUDA0 compute buffer size = 2.77 MiB
alloc_compute_meta: CPU compute buffer size = 0.16 MiB
srv load_model: loaded multimodal model, '/models/Qwen2.5-VL-7B-Instruct.mmproj-fp16.gguf'
srv load_model: ctx_shift is not supported by multimodal, it will be disabled
srv init: initializing slots, n_slots = 1
slot init: id 0 | task -1 | new slot n_ctx_slot = 4096
main: model loaded
main: chat template, chat_template: {%- if tools %}
{{- '<|im_start|>system\n' }}
{%- if messages[0]['role'] == 'system' %}
{{- messages[0]['content'] }}
{%- else %}
{{- 'You are a helpful assistant.' }}
{%- endif %}
{{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n" }}
{%- for tool in tools %}
{{- "\n" }}
{{- tool | tojson }}
{%- endfor %}
{{- "\n\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{"name": , "arguments": }\n</tool_call><|im_end|>\n" }}
{%- else %}
{%- if messages[0]['role'] == 'system' %}
{{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
{%- else %}
{{- '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}
{%- endif %}
�
{%- endif %}
{%- for message in messages %}
{%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
{{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
{%- elif message.role == "assistant" %}
{{- '<|im_start|>' + message.role }}
{%- if message.content %}
{{- '\n' + message.content }}
{%- endif %}
{%- for tool_call in message.tool_calls %}
{%- if tool_call.function is defined %}
{%- set tool_call = tool_call.function %}
{%- endif %}
{{- '\n<tool_call>\n{"name": "' }}
{{- tool_call.name }}
{{- '", "arguments": ' }}
{{- tool_call.arguments | tojson }}
{{- '}\n</tool_call>' }}
{%- endfor %}
{{- '<|im_end|>\n' }}
{%- elif message.role == "tool" %}
{%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
{{- '<|im_start|>user' }}
{%- endif %}
{{- '\n<tool_response>\n' }}
{{- message.content }}
{{- '\n</tool_response>' }}
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
{{- '<|im_end|>\n' }}
{%- endif %}
{%- endif %}
{%- endfor %}
{%- if add_generation_prompt %}
{{- '<|im_start|>assistant\n' }}
�
{%- 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://0.0.0.0:8080 - starting the main loop
srv update_slots: all slots are idle
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /v1/api/tags 172.23.0.1 404
srv log_server_r: request: GET /v1/api/version 172.23.0.1 404
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET / 192.168.2.103 200
srv log_server_r: request: GET /v1/api/tags 172.23.0.1 404
srv log_server_r: request: GET /favicon.ico 192.168.2.103 404
srv log_server_r: request: GET /props 192.168.2.103 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv log_server_r: request: GET /health 127.0.0.1 200
srv params_from_: Chat format: Content-only
slot launch_slot_: id 0 | task 0 | processing task
slot update_slots: id 0 | task 0 | new prompt, n_ctx_slot = 4096, n_keep = 0, n_prompt_tokens = 14
slot update_slots: id 0 | task 0 | kv cache rm [0, end)
slot update_slots: id 0 | task 0 | prompt processing progress, n_past = 4, n_tokens = 4, progress = 0.285714
slot update_slots: id 0 | task 0 | kv cache rm [4, end)
srv process_chun: processing image...
ggml_backend_cuda_buffer_type_alloc_buffer: allocating 8641.89 MiB on device 0: cudaMalloc failed: out of memory
ggml_gallocr_reserve_n: failed to allocate CUDA0 buffer of size 9061680896
/app/ggml/src/ggml-backend.cpp:1666: GGML_ASSERT((char *)addr + ggml_backend_buffer_get_alloc_size(buffer, tensor) <= (char *)ggml_backend_buffer_get_base(buffer) + ggml_backend_buffer_get_size(buffer)) failed
encoding image or slice...

[domasofan]

Just tried at home with build 5359 and this model:
https://huggingface.co/ggml-org/Qwen2.5-VL-7B-Instruct-GGUF
Seems to work again. will try tomorrow in the office where i have a different model.

[domasofan]

Works as well here in the office.
Used build: b5361

[zandera007]

Ok,
il will try this evening with the last build B5361 (i'm using docker).

I don't understand with they are so much memory allocated in the buffer when the model receive the image... I hope that the last build fix this issue.

[zandera007]

I confirm, it's working with the latest build.