Eval bug: GGML_ASSERT(ggml_is_contiguous(a)) with Jina reranker model #15895
Description
Name and Version
.\llama-server.exe --version
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon RX 6600M (AMD proprietary driver) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 32768 | int dot: 1 | matrix cores: none
version: 6428 (a972fae)
built with MSVC 19.38.33134.0 for x64
Operating systems
Windows
GGML backends
Vulkan
Hardware
Ryzen 7 5800H + AMD Radeon RX 6600M
Models
jina-reranker-v2-base-multilingual
Problem description & steps to reproduce
When I try to serve the Jina reranker model, it throws C:\Sources\llama.cpp\ggml\src\ggml.c:3435: GGML_ASSERT(ggml_is_contiguous(a)) failed error. Tried enabling/disabling FA, changing context/batch size with no success as well.
First Bad Commit
No response
Relevant log output
PS C:\Sources\llama.cpp\build\bin\Release> .\llama-server.exe --reranking -ub 1024 -b 1024 -c 1024 --host 192.168.100.21 --port 8081 -m C:\Temp\Jina-Reranker-v2-Base-Multilingual-278M-Q8_0.gguf -ngl 99 -fa 0
ggml_vulkan: Found 1 Vulkan devices:
ggml_vulkan: 0 = AMD Radeon RX 6600M (AMD proprietary driver) | uma: 0 | fp16: 1 | bf16: 0 | warp size: 32 | shared memory: 32768 | int dot: 1 | matrix cores: none
build: 6428 (a972faebe) with MSVC 19.38.33134.0 for x64
system info: n_threads = 8, n_threads_batch = 8, total_threads = 16
system_info: n_threads = 8 (n_threads_batch = 8) / 16 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 |
main: binding port with default address family
main: HTTP server is listening, hostname: 192.168.100.21, port: 8081, http threads: 15
main: loading model
srv load_model: loading model 'C:\Temp\Jina-Reranker-v2-Base-Multilingual-278M-Q8_0.gguf'
llama_model_load_from_file_impl: using device Vulkan0 (AMD Radeon RX 6600M) - 7360 MiB free
llama_model_loader: loaded meta data with 35 key-value pairs and 153 tensors from C:\Temp\Jina-Reranker-v2-Base-Multilingual-278M-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 = bert
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.name str = Jina Reranker v2 Base Multilingual
llama_model_loader: - kv 3: general.organization str = Jinaai
llama_model_loader: - kv 4: general.size_label str = 278M
llama_model_loader: - kv 5: general.license str = cc-by-nc-4.0
llama_model_loader: - kv 6: general.tags arr[str,5] = ["transformers", "reranker", "cross-e...
llama_model_loader: - kv 7: general.languages arr[str,1] = ["multilingual"]
llama_model_loader: - kv 8: bert.block_count u32 = 12
llama_model_loader: - kv 9: bert.context_length u32 = 1024
llama_model_loader: - kv 10: bert.embedding_length u32 = 768
llama_model_loader: - kv 11: bert.feed_forward_length u32 = 3072
llama_model_loader: - kv 12: bert.attention.head_count u32 = 12
llama_model_loader: - kv 13: bert.attention.layer_norm_epsilon f32 = 0.000010
llama_model_loader: - kv 14: general.file_type u32 = 7
llama_model_loader: - kv 15: bert.attention.causal bool = false
llama_model_loader: - kv 16: bert.classifier.output_labels arr[str,1] = ["LABEL_0"]
llama_model_loader: - kv 17: general.quantization_version u32 = 2
llama_model_loader: - kv 18: tokenizer.ggml.model str = t5
llama_model_loader: - kv 19: tokenizer.ggml.pre str = default
llama_model_loader: - kv 20: tokenizer.ggml.tokens arr[str,250002] = ["<s>", "<pad>", "</s>", "<unk>", ","...
llama_model_loader: - kv 21: tokenizer.ggml.scores arr[f32,250002] = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv 22: tokenizer.ggml.token_type arr[i32,250002] = [3, 3, 3, 2, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv 23: tokenizer.ggml.add_space_prefix bool = false
llama_model_loader: - kv 24: tokenizer.ggml.token_type_count u32 = 1
llama_model_loader: - kv 25: tokenizer.ggml.remove_extra_whitespaces bool = true
llama_model_loader: - kv 26: tokenizer.ggml.precompiled_charsmap arr[u8,237539] = [0, 180, 2, 0, 0, 132, 0, 0, 0, 0, 0,...
llama_model_loader: - kv 27: tokenizer.ggml.bos_token_id u32 = 0
llama_model_loader: - kv 28: tokenizer.ggml.eos_token_id u32 = 2
llama_model_loader: - kv 29: tokenizer.ggml.unknown_token_id u32 = 3
llama_model_loader: - kv 30: tokenizer.ggml.seperator_token_id u32 = 2
llama_model_loader: - kv 31: tokenizer.ggml.padding_token_id u32 = 1
llama_model_loader: - kv 32: tokenizer.ggml.mask_token_id u32 = 250001
llama_model_loader: - kv 33: tokenizer.ggml.add_bos_token bool = true
llama_model_loader: - kv 34: tokenizer.ggml.add_eos_token bool = true
llama_model_loader: - type f32: 102 tensors
llama_model_loader: - type q8_0: 51 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q8_0
print_info: file size = 284.68 MiB (8.58 BPW)
load: model vocab missing newline token, using special_pad_id instead
load: special_eos_id is not in special_eog_ids - the tokenizer config may be incorrect
load: printing all EOG tokens:
load: - 2 ('</s>')
load: special tokens cache size = 5
load: token to piece cache size = 2.1668 MB
print_info: arch = bert
print_info: vocab_only = 0
print_info: n_ctx_train = 1024
print_info: n_embd = 768
print_info: n_layer = 12
print_info: n_head = 12
print_info: n_head_kv = 12
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 = 1
print_info: n_embd_k_gqa = 768
print_info: n_embd_v_gqa = 768
print_info: f_norm_eps = 1.0e-05
print_info: f_norm_rms_eps = 0.0e+00
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 = 3072
print_info: n_expert = 0
print_info: n_expert_used = 0
print_info: causal attn = 0
print_info: pooling type = -1
print_info: rope type = 2
print_info: rope scaling = linear
print_info: freq_base_train = 10000.0
print_info: freq_scale_train = 1
print_info: n_ctx_orig_yarn = 1024
print_info: rope_finetuned = unknown
print_info: n_cls_out = 1
print_info: cls_label[ 0] = LABEL_0
print_info: model type = 109M
print_info: model params = 278.44 M
print_info: general.name = Jina Reranker v2 Base Multilingual
print_info: vocab type = UGM
print_info: n_vocab = 250002
print_info: n_merges = 0
print_info: BOS token = 0 '<s>'
print_info: EOS token = 2 '</s>'
print_info: UNK token = 3 '<unk>'
print_info: SEP token = 2 '</s>'
print_info: PAD token = 1 '<pad>'
print_info: MASK token = 250001 '<mask>'
print_info: LF token = 0 '<s>'
print_info: EOG token = 2 '</s>'
print_info: max token length = 48
load_tensors: loading model tensors, this can take a while... (mmap = true)
load_tensors: offloading 12 repeating layers to GPU
load_tensors: offloading output layer to GPU
load_tensors: offloaded 13/13 layers to GPU
load_tensors: Vulkan0 model buffer size = 87.12 MiB
load_tensors: CPU_Mapped model buffer size = 197.56 MiB
.................................
llama_context: constructing llama_context
llama_context: n_seq_max = 1
llama_context: n_ctx = 1024
llama_context: n_ctx_per_seq = 1024
llama_context: n_batch = 1024
llama_context: n_ubatch = 1024
llama_context: causal_attn = 0
llama_context: flash_attn = disabled
llama_context: kv_unified = false
llama_context: freq_base = 10000.0
llama_context: freq_scale = 1
llama_context: Vulkan_Host output buffer size = 0.96 MiB
C:\Sources\llama.cpp\ggml\src\ggml.c:3435: GGML_ASSERT(ggml_is_contiguous(a)) failed