Eval bug: Crash on first message in llama.cpp with CUDA backend when running via docker #17076
Description
Name and Version
$ docker run --rm -it -v /home/dusko/Documents/Models/LLM:/models -p 8000:8000 --device nvidia.com/gpu=0 ghcr.io/ggml-org/llama.cpp:server-cuda -m /models/unsloth/gemma-3n-E2B-it-GGUF/gemma-3n-E2B-it-Q4_0.gguf --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 RTX 5060 Ti, compute capability 12.0, VMM: yes
load_backend: loaded CUDA backend from /app/libggml-cuda.so
load_backend: loaded CPU backend from /app/libggml-cpu-icelake.so
version: 6970 (7f09a680a)
built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnuOperating systems
Linux
GGML backends
CUDA
Hardware
GPU: RTX 5060 Ti 16 GB
Models
unsloth/gemma-3n-E2B-it-Q4_0.gguf
Problem description & steps to reproduce
Start the llama.cpp using docker image compiled with cuda:
$ docker run --rm -it -v /home/dusko/Documents/Models/LLM:/models -p 8000:8000 --device nvidia.com/gpu=0 ghcr.io/ggml-org/llama.cpp:server-cuda -m /models/unsloth/gemma-3n-E2B-it-GGUF/gemma-3n-E2B-it-Q4_0.gguf --port 8000 -c 1024 -ngl 999 -fa on --jinjaSend a Hi message using the llama.cpp web interface and it will do some processing for a bit before crashing. I was able to run everything normally, without any issues, before updating to latest docker image.
First Bad Commit
Latest ghcr.io/ggml-org/llama.cpp:server-cuda docker image
git is not available from inside of the docker image so I was not able to verify
Relevant log output
$ docker run --rm -it -v /home/dusko/Documents/Models/LLM:/models -p 8000:8000 --device nvidia.com/gpu=0 ghcr.io/ggml-org/llama.cpp:server-cuda -m /models/unsloth/gemma-3n-E2B-it-GGUF/gemma-3n-E2B-it-Q4_0.gguf --port 8000 -c 1024 -ngl 999 -fa on --jinja
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 RTX 5060 Ti, compute capability 12.0, VMM: yes
load_backend: loaded CUDA backend from /app/libggml-cuda.so
load_backend: loaded CPU backend from /app/libggml-cpu-icelake.so
main: setting n_parallel = 4 and kv_unified = true (add -kvu to disable this)
build: 6970 (7f09a680a) with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
system info: n_threads = 12, n_threads_batch = 12, total_threads = 24
system_info: n_threads = 12 (n_threads_batch = 12) / 24 | 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 | AVX512 = 1 | AVX512_VBMI = 1 | AVX512_VNNI = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 |
main: binding port with default address family
main: HTTP server is listening, hostname: 0.0.0.0, port: 8000, http threads: 23
main: loading model
srv load_model: loading model '/models/unsloth/gemma-3n-E2B-it-GGUF/gemma-3n-E2B-it-Q4_0.gguf'
llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce RTX 5060 Ti) (0000:01:00.0) - 15699 MiB free
llama_model_loader: loaded meta data with 51 key-value pairs and 727 tensors from /models/unsloth/gemma-3n-E2B-it-GGUF/gemma-3n-E2B-it-Q4_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 = gemma3n
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.name str = Gemma-3N-E2B-It
llama_model_loader: - kv 3: general.finetune str = 3n-E2B-it
llama_model_loader: - kv 4: general.basename str = Gemma-3N-E2B-It
llama_model_loader: - kv 5: general.quantized_by str = Unsloth
llama_model_loader: - kv 6: general.size_label str = 4.5B
llama_model_loader: - kv 7: general.license str = gemma
llama_model_loader: - kv 8: general.repo_url str = https://huggingface.co/unsloth
llama_model_loader: - kv 9: general.base_model.count u32 = 1
llama_model_loader: - kv 10: general.base_model.0.name str = Gemma 3n E2B It
llama_model_loader: - kv 11: general.base_model.0.organization str = Google
llama_model_loader: - kv 12: general.base_model.0.repo_url str = https://huggingface.co/google/gemma-3...
llama_model_loader: - kv 13: general.tags arr[str,6] = ["automatic-speech-recognition", "uns...
llama_model_loader: - kv 14: gemma3n.context_length u32 = 32768
llama_model_loader: - kv 15: gemma3n.embedding_length u32 = 2048
llama_model_loader: - kv 16: gemma3n.block_count u32 = 30
llama_model_loader: - kv 17: gemma3n.feed_forward_length arr[i32,30] = [8192, 8192, 8192, 8192, 8192, 8192, ...
llama_model_loader: - kv 18: gemma3n.attention.head_count u32 = 8
llama_model_loader: - kv 19: gemma3n.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 20: gemma3n.attention.key_length u32 = 256
llama_model_loader: - kv 21: gemma3n.attention.value_length u32 = 256
llama_model_loader: - kv 22: gemma3n.rope.freq_base f32 = 1000000.000000
llama_model_loader: - kv 23: gemma3n.attention.sliding_window u32 = 512
llama_model_loader: - kv 24: gemma3n.attention.head_count_kv u32 = 2
llama_model_loader: - kv 25: gemma3n.altup.active_idx u32 = 0
llama_model_loader: - kv 26: gemma3n.altup.num_inputs u32 = 4
llama_model_loader: - kv 27: gemma3n.embedding_length_per_layer_input u32 = 256
llama_model_loader: - kv 28: gemma3n.attention.shared_kv_layers u32 = 10
llama_model_loader: - kv 29: gemma3n.activation_sparsity_scale arr[f32,30] = [1.644854, 1.644854, 1.644854, 1.6448...
llama_model_loader: - kv 30: gemma3n.attention.sliding_window_pattern arr[bool,30] = [true, true, true, true, false, true,...
llama_model_loader: - kv 31: tokenizer.chat_template str = {{ bos_token }}\n{%- if messages[0]['r...
llama_model_loader: - kv 32: tokenizer.ggml.model str = llama
llama_model_loader: - kv 33: tokenizer.ggml.pre str = default
llama_model_loader: - kv 34: tokenizer.ggml.tokens arr[str,262144] = ["<pad>", "<eos>", "<bos>", "<unk>", ...
llama_model_loader: - kv 35: tokenizer.ggml.scores arr[f32,262144] = [-1000.000000, -1000.000000, -1000.00...
llama_model_loader: - kv 36: tokenizer.ggml.token_type arr[i32,262144] = [3, 3, 3, 3, 3, 4, 3, 3, 3, 3, 3, 3, ...
llama_model_loader: - kv 37: tokenizer.ggml.bos_token_id u32 = 2
llama_model_loader: - kv 38: tokenizer.ggml.eos_token_id u32 = 106
llama_model_loader: - kv 39: tokenizer.ggml.unknown_token_id u32 = 3
llama_model_loader: - kv 40: tokenizer.ggml.padding_token_id u32 = 0
llama_model_loader: - kv 41: tokenizer.ggml.add_bos_token bool = true
llama_model_loader: - kv 42: tokenizer.ggml.add_sep_token bool = false
llama_model_loader: - kv 43: tokenizer.ggml.add_eos_token bool = false
llama_model_loader: - kv 44: tokenizer.ggml.add_space_prefix bool = false
llama_model_loader: - kv 45: general.quantization_version u32 = 2
llama_model_loader: - kv 46: general.file_type u32 = 2
llama_model_loader: - kv 47: quantize.imatrix.file str = gemma-3n-E2B-it-GGUF/imatrix_unsloth.dat
llama_model_loader: - kv 48: quantize.imatrix.dataset str = unsloth_calibration_gemma-3n-E2B-it.txt
llama_model_loader: - kv 49: quantize.imatrix.entries_count u32 = 400
llama_model_loader: - kv 50: quantize.imatrix.chunks_count u32 = 1326
llama_model_loader: - type f32: 362 tensors
llama_model_loader: - type f16: 93 tensors
llama_model_loader: - type q4_0: 267 tensors
llama_model_loader: - type q4_1: 3 tensors
llama_model_loader: - type q5_1: 1 tensors
llama_model_loader: - type q4_K: 1 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q4_0
print_info: file size = 2.76 GiB (5.31 BPW)
load: printing all EOG tokens:
load: - 106 ('<end_of_turn>')
load: special tokens cache size = 6414
load: token to piece cache size = 1.9446 MB
print_info: arch = gemma3n
print_info: vocab_only = 0
print_info: n_ctx_train = 32768
print_info: n_embd = 2048
print_info: n_layer = 30
print_info: n_head = 8
print_info: n_head_kv = 2
print_info: n_rot = 256
print_info: n_swa = 512
print_info: is_swa_any = 1
print_info: n_embd_head_k = 256
print_info: n_embd_head_v = 256
print_info: n_gqa = 4
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 = 1.0e+00
print_info: n_ff = 8192
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 = 32768
print_info: rope_finetuned = unknown
print_info: model type = E2B
print_info: model params = 4.46 B
print_info: general.name = Gemma-3N-E2B-It
print_info: vocab type = SPM
print_info: n_vocab = 262144
print_info: n_merges = 0
print_info: BOS token = 2 '<bos>'
print_info: EOS token = 106 '<end_of_turn>'
print_info: EOT token = 106 '<end_of_turn>'
print_info: UNK token = 3 '<unk>'
print_info: PAD token = 0 '<pad>'
print_info: LF token = 248 '<0x0A>'
print_info: EOG token = 106 '<end_of_turn>'
print_info: max token length = 48
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 = 288.00 MiB
load_tensors: CUDA0 model buffer size = 2822.05 MiB
.......................................
llama_context: constructing llama_context
llama_context: n_seq_max = 4
llama_context: n_ctx = 1024
llama_context: n_ctx_seq = 1024
llama_context: n_batch = 1024
llama_context: n_ubatch = 512
llama_context: causal_attn = 1
llama_context: flash_attn = enabled
llama_context: kv_unified = true
llama_context: freq_base = 1000000.0
llama_context: freq_scale = 1
llama_context: n_ctx_seq (1024) < n_ctx_train (32768) -- the full capacity of the model will not be utilized
llama_context: CUDA_Host output buffer size = 4.00 MiB
llama_kv_cache_iswa: creating non-SWA KV cache, size = 1024 cells
llama_kv_cache: CUDA0 KV buffer size = 8.00 MiB
llama_kv_cache: size = 8.00 MiB ( 1024 cells, 4 layers, 4/1 seqs), K (f16): 4.00 MiB, V (f16): 4.00 MiB
llama_kv_cache_iswa: creating SWA KV cache, size = 1024 cells
llama_kv_cache: CUDA0 KV buffer size = 32.00 MiB
llama_kv_cache: size = 32.00 MiB ( 1024 cells, 16 layers, 4/1 seqs), K (f16): 16.00 MiB, V (f16): 16.00 MiB
llama_context: CUDA0 compute buffer size = 520.00 MiB
llama_context: CUDA_Host compute buffer size = 8.02 MiB
llama_context: graph nodes = 2733
llama_context: graph splits = 2
common_init_from_params: added <end_of_turn> logit bias = -inf
common_init_from_params: setting dry_penalty_last_n to ctx_size = 1024
common_init_from_params: warming up the model with an empty run - please wait ... (--no-warmup to disable)
srv init: initializing slots, n_slots = 4
slot init: id 0 | task -1 | new slot, n_ctx = 1024
slot init: id 1 | task -1 | new slot, n_ctx = 1024
slot init: id 2 | task -1 | new slot, n_ctx = 1024
slot init: id 3 | task -1 | new slot, n_ctx = 1024
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 }}
{%- if messages[0]['role'] == 'system' -%}
{%- if messages[0]['content'] is string -%}
{%- set first_user_prefix = messages[0]['content'] + '
' -%}
{%- else -%}
{%- set first_user_prefix = messages[0]['content'][0]['text'] + '
' -%}
{%- endif -%}
{%- set loop_messages = messages[1:] -%}
{%- else -%}
{%- set first_user_prefix = "" -%}
{%- set loop_messages = messages -%}
{%- endif -%}
{%- for message in loop_messages -%}
{%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
{{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
{%- endif -%}
{%- if (message['role'] == 'assistant') -%}
{%- set role = "model" -%}
{%- else -%}
{%- set role = message['role'] -%}
{%- endif -%}
{{ '<start_of_turn>' + role + '
' + (first_user_prefix if loop.first else "") }}
{%- if message['content'] is string -%}
{{ message['content'] | trim }}
{%- elif message['content'] is iterable -%}
{%- for item in message['content'] -%}
{%- if item['type'] == 'audio' -%}
{{ '<audio_soft_token>' }}
{%- elif item['type'] == 'image' -%}
{{ '<image_soft_token>' }}
{%- elif item['type'] == 'text' -%}
{{ item['text'] | trim }}
{%- endif -%}
{%- endfor -%}
{%- else -%}
{{ raise_exception("Invalid content type") }}
{%- endif -%}
{{ '<end_of_turn>
' }}
{%- endfor -%}
{%- if add_generation_prompt -%}
{{'<start_of_turn>model
'}}
{%- endif -%}
, example_format: '<start_of_turn>user
You are a helpful assistant
Hello<end_of_turn>
<start_of_turn>model
Hi there<end_of_turn>
<start_of_turn>user
How are you?<end_of_turn>
<start_of_turn>model
'
main: server is listening on http://0.0.0.0:8000 - starting the main loop
srv update_slots: all slots are idle
srv log_server_r: request: GET / 192.168.1.5 200
srv log_server_r: request: GET /props 192.168.1.5 200
srv log_server_r: request: GET /props 192.168.1.5 200
srv log_server_r: request: GET /props 192.168.1.5 200
srv update_slots: all slots are idle
srv log_server_r: request: GET /slots 192.168.1.5 200
srv log_server_r: request: GET /props 192.168.1.5 200
srv log_server_r: request: GET /props 192.168.1.5 200
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 1 | processing task
slot update_slots: id 3 | task 1 | new prompt, n_ctx_slot = 1024, n_keep = 0, task.n_tokens = 10
slot update_slots: id 3 | task 1 | n_tokens = 0, memory_seq_rm [0, end)
slot update_slots: id 3 | task 1 | prompt processing progress, n_tokens = 10, batch.n_tokens = 10, progress = 1.000000
slot update_slots: id 3 | task 1 | prompt done, n_tokens = 10, batch.n_tokens = 10
srv log_server_r: request: GET /props 192.168.1.5 200
/app/ggml/src/ggml-cuda/cpy.cu:210: GGML_ASSERT(false) failed
libggml-base.so(+0x183cb)[0x7784537fa3cb]
libggml-base.so(ggml_print_backtrace+0x21f)[0x7784537fa82f]
libggml-base.so(ggml_abort+0x152)[0x7784537faa02]
/app/libggml-cuda.so(+0xed380)[0x77843a667380]
/app/libggml-cuda.so(_Z13ggml_cuda_cpyR25ggml_backend_cuda_contextPK11ggml_tensorPS1_+0x3bc)[0x77843a66e3fc]
/app/libggml-cuda.so(+0x13fe52)[0x77843a6b9e52]
/app/libggml-cuda.so(+0x141aef)[0x77843a6bbaef]
libggml-base.so(ggml_backend_sched_graph_compute_async+0x81f)[0x77845381695f]
libllama.so(_ZN13llama_context13graph_computeEP11ggml_cgraphb+0xa1)[0x778453937581]
libllama.so(_ZN13llama_context14process_ubatchERK12llama_ubatch14llm_graph_typeP22llama_memory_context_iR11ggml_status+0x11d)[0x77845393792d]
libllama.so(_ZN13llama_context6decodeERK11llama_batch+0x2e7)[0x77845393e2d7]
libllama.so(llama_decode+0x10)[0x77845393f170]
/app/llama-server(+0xf7748)[0x5c1b3fa59748]
/app/llama-server(+0x98cd8)[0x5c1b3f9facd8]
/app/llama-server(+0x57bcd)[0x5c1b3f9b9bcd]
/usr/lib/x86_64-linux-gnu/libc.so.6(+0x29d90)[0x7784532add90]
/usr/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0x80)[0x7784532ade40]
/app/llama-server(+0x59685)[0x5c1b3f9bb685]