Eval bug: missing tensor 'blk.40.ssm_conv1d.weight'

[IAMJOYBO]

Name and Version

C:\Users\13268>llama-cli --version
ggml_cuda_init: found 2 CUDA devices (Total VRAM: 28663 MiB):
  Device 0: NVIDIA GeForce RTX 4080 SUPER, compute capability 8.9, VMM: yes, VRAM: 16375 MiB
  Device 1: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes, VRAM: 12287 MiB
load_backend: loaded CUDA backend from D:\USER_DATA\Documents\Software\llama.cpp\ggml-cuda.dll
load_backend: loaded RPC backend from D:\USER_DATA\Documents\Software\llama.cpp\ggml-rpc.dll
load_backend: loaded CPU backend from D:\USER_DATA\Documents\Software\llama.cpp\ggml-cpu-alderlake.dll
version: 9143 (7f3f843c3)
built with Clang 19.1.5 for Windows x86_64

C:\Users\13268>

Operating systems

Windows

GGML backends

CUDA

Hardware

Product Name: MKT-MPG Z790 EDGE WIFI
OS: Microsoft Windows 11 专业工作站版 64 位 Ver.2009 (OS build 26200.8457)
CPU: 12th Gen Intel(R) Core(TM) i5-12600KF
Memory: 64 GB @
- 32 GB DDR5-4800, Corsair CMK5X32G2B56C36A2
- 32 GB DDR5-4800, Corsair CMK5X32G2B56C36A2
Graphics: NVIDIA GeForce RTX 4080 SUPER
Graphics: RTX 3060 GAMING (X/Z) TRIO 12G, 12288 MB

Models

unsloth\Qwen3.6-35B-A3B-MTP-GGUF\Qwen3.6-35B-A3B-UD-IQ4_NL.gguf

Problem description & steps to reproduce

使用 llama-server.exe 启动

First Bad Commit

No response

Relevant log output

Logs

llama.cpp version output: ggml_cuda_init: found 2 CUDA devices (Total VRAM: 28663 MiB):
  Device 0: NVIDIA GeForce RTX 4080 SUPER, compute capability 8.9, VMM: yes, VRAM: 16375 MiB
  Device 1: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes, VRAM: 12287 MiB
load_backend: loaded CUDA backend from C:\Users\13268\llama.cpp\ggml-cuda.dll
load_backend: loaded RPC backend from C:\Users\13268\llama.cpp\ggml-rpc.dll
load_backend: loaded CPU backend from C:\Users\13268\llama.cpp\ggml-cpu-alderlake.dll
version: 9143 (7f3f843c3)
built with Clang 19.1.5 for Windows x86_64
Extracted version: 9143
llama.cpp version output: ggml_cuda_init: found 2 CUDA devices (Total VRAM: 28663 MiB):
  Device 0: NVIDIA GeForce RTX 4080 SUPER, compute capability 8.9, VMM: yes, VRAM: 16375 MiB
  Device 1: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes, VRAM: 12287 MiB
load_backend: loaded CUDA backend from C:\Users\13268\llama.cpp\ggml-cuda.dll
load_backend: loaded RPC backend from C:\Users\13268\llama.cpp\ggml-rpc.dll
load_backend: loaded CPU backend from C:\Users\13268\llama.cpp\ggml-cpu-alderlake.dll
version: 9143 (7f3f843c3)
built with Clang 19.1.5 for Windows x86_64
Extracted version: 9143
Starting llama.cpp server with args: [
  '--host',
  '127.0.0.1',
  '--port',
  '8080',
  '--api-key',
  '1234567890',
  '--ctx-size',
  '200000',
  '--gpu-layers',
  '999',
  '--flash-attn',
  'on',
  '--reasoning',
  'off',
  '--threads',
  '8',
  '--model',
  'E:\\llm-model\\unsloth\\Qwen3.6-35B-A3B-MTP-GGUF\\Qwen3.6-35B-A3B-UD-IQ4_NL.gguf'
]
llama-server stderr: ggml_cuda_init: found 2 CUDA devices (Total VRAM: 28663 MiB):
  Device 0: NVIDIA GeForce RTX 4080 SUPER, compute capability 8.9, VMM: yes, VRAM: 16375 MiB
  Device 1: NVIDIA GeForce RTX 3060, compute capability 8.6, VMM: yes, VRAM: 12287 MiB
load_backend: loaded CUDA backend from C:\Users\13268\llama.cpp\ggml-cuda.dll

llama-server stderr: load_backend: loaded RPC backend from C:\Users\13268\llama.cpp\ggml-rpc.dll

llama-server stderr: load_backend: loaded CPU backend from C:\Users\13268\llama.cpp\ggml-cpu-alderlake.dll

llama-server stderr: main: n_parallel is set to auto, using n_parallel = 4 and kv_unified = true
build_info: b9143-7f3f843c3

llama-server stderr: system_info: n_threads = 8 (n_threads_batch = 8) / 16 | CUDA : ARCHS = 750,800,860,890,1200,1210 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX_VNNI = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 |
Running without SSL
init: api_keys: ****7890
init: using 15 threads for HTTP server
start: binding port with default address family

llama-server stderr: main: loading model
srv    load_model: loading model 'E:\llm-model\unsloth\Qwen3.6-35B-A3B-MTP-GGUF\Qwen3.6-35B-A3B-UD-IQ4_NL.gguf'
common_init_result: fitting params to device memory, for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on
common_params_fit_impl: getting device memory data for initial parameters:

llama-server stderr: llama_model_load: error loading model: missing tensor 'blk.40.ssm_conv1d.weight'
llama_model_load_from_file_impl: failed to load model
common_fit_params: encountered an error while trying to fit params to free device memory: failed to load model
common_fit_params: fitting params to free memory took 0.60 seconds

llama-server stderr: llama_model_loader: loaded meta data with 55 key-value pairs and 753 tensors from E:\llm-model\unsloth\Qwen3.6-35B-A3B-MTP-GGUF\Qwen3.6-35B-A3B-UD-IQ4_NL.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              = qwen35moe
llama_model_loader: - kv   1:                               general.type str              = model
llama_model_loader: - kv   2:                     general.sampling.top_k i32              = 20
llama_model_loader: - kv   3:                     general.sampling.top_p f32              = 0.950000
llama_model_loader: - kv   4:                      general.sampling.temp f32              = 1.000000
llama_model_loader: - kv   5:                               general.name str              = Qwen3.6-35B-A3B
llama_model_loader: - kv   6:                           general.basename str              = Qwen3.6-35B-A3B
llama_model_loader: - kv   7:                       general.quantized_by str              = Unsloth
llama_model_loader: - kv   8:                         general.size_label str              = 35B-A3B
llama_model_loader: - kv   9:                            general.license str              = apache-2.0
llama_model_loader: - kv  10:                       general.license.link str              = https://huggingface.co/Qwen/Qwen3.6-3...
llama_model_loader: - kv  11:                           general.repo_url str              = https://huggingface.co/unsloth
llama_model_loader: - kv  12:                   general.base_model.count u32              = 1
llama_model_loader: - kv  13:                  general.base_model.0.name str              = Qwen3.6 35B A3B
llama_model_loader: - kv  14:          general.base_model.0.organization str              = Qwen
llama_model_loader: - kv  15:              general.base_model.0.repo_url str              = https://huggingface.co/Qwen/Qwen3.6-3...
llama_model_loader: - kv  16:                               general.tags arr[str,3]       = ["qwen3_5_moe", "qwen", "image-text-t...
llama_model_loader: - kv  17:                      qwen35moe.block_count u32              = 41
llama_model_loader: - kv  18:                   qwen35moe.context_length u32              = 262144
llama_model_loader: - kv  19:                 qwen35moe.embedding_length u32              = 2048
llama_model_loader: - kv  20:             qwen35moe.attention.head_count u32              = 16
llama_model_loader: - kv  21:          qwen35moe.attention.head_count_kv u32              = 2
llama_model_loader: - kv  22:          qwen35moe.rope.dimension_sections arr[i32,4]       = [11, 11, 10, 0]
llama_model_loader: - kv  23:                   qwen35moe.rope.freq_base f32              = 10000000.000000
llama_model_loader: - kv  24: qwen35moe.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv  25:                     qwen35moe.expert_count u32              = 256
llama_model_loader: - kv  26:                qwen35moe.expert_used_count u32              = 8
llama_model_loader: - kv  27:             qwen35moe.attention.key_length u32              = 256
llama_model_loader: - kv  28:           qwen35moe.attention.value_length u32              = 256
llama_model_loader: - kv  29:       qwen35moe.expert_feed_forward_length u32              = 512
llama_model_loader: - kv  30: qwen35moe.expert_shared_feed_forward_length u32              = 512
llama_model_loader: - kv  31:                  qwen35moe.ssm.conv_kernel u32              = 4
llama_model_loader: - kv  32:                   qwen35moe.ssm.state_size u32              = 128
llama_model_loader: - kv  33:                  qwen35moe.ssm.group_count u32              = 16
llama_model_loader: - kv  34:               qwen35moe.ssm.time_step_rank u32              = 32
llama_model_loader: - kv  35:                   qwen35moe.ssm.inner_size u32              = 4096
llama_model_loader: - kv  36:          qwen35moe.full_attention_interval u32              = 4
llama_model_loader: - kv  37:             qwen35moe.rope.dimension_count u32              = 64
llama_model_loader: - kv  38:             qwen35moe.nextn_predict_layers u32              = 1
llama_model_loader: - kv  39:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  40:                         tokenizer.ggml.pre str              = qwen35

llama-server stderr: llama_model_loader: - kv  41:                      tokenizer.ggml.tokens arr[str,248320]  = ["!", "\"", "#", "$", "%", "&", "'", ...

llama-server stderr: llama_model_loader: - kv  42:                  tokenizer.ggml.token_type arr[i32,248320]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...

llama-server stderr: llama_model_loader: - kv  43:                      tokenizer.ggml.merges arr[str,247587]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv  44:                tokenizer.ggml.eos_token_id u32              = 248046
llama_model_loader: - kv  45:            tokenizer.ggml.padding_token_id u32              = 248055
llama_model_loader: - kv  46:                tokenizer.ggml.bos_token_id u32              = 248044
llama_model_loader: - kv  47:               tokenizer.ggml.add_bos_token bool             = false
llama_model_loader: - kv  48:                    tokenizer.chat_template str              = {%- set image_count = namespace(value...
llama_model_loader: - kv  49:               general.quantization_version u32              = 2
llama_model_loader: - kv  50:                          general.file_type u32              = 25
llama_model_loader: - kv  51:                      quantize.imatrix.file str              = Qwen3.6-35B-A3B-GGUF/imatrix_unsloth....
llama_model_loader: - kv  52:                   quantize.imatrix.dataset str              = unsloth_calibration_Qwen3.6-35B-A3B.txt
llama_model_loader: - kv  53:             quantize.imatrix.entries_count u32              = 510
llama_model_loader: - kv  54:              quantize.imatrix.chunks_count u32              = 77
llama_model_loader: - type  f32:  368 tensors
llama_model_loader: - type q8_0:  259 tensors
llama_model_loader: - type q3_K:    2 tensors
llama_model_loader: - type q4_K:    1 tensors
llama_model_loader: - type q6_K:    4 tensors
llama_model_loader: - type iq4_nl:   39 tensors
llama_model_loader: - type iq3_s:   78 tensors
llama_model_loader: - type bf16:    2 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type   = IQ4_NL - 4.5 bpw
print_info: file size   = 17.25 GiB (4.17 BPW)
llama_prepare_model_devices: using device CUDA0 (NVIDIA GeForce RTX 4080 SUPER) (0000:01:00.0) - 15061 MiB free
llama_prepare_model_devices: using device CUDA1 (NVIDIA GeForce RTX 3060) (0000:08:00.0) - 11253 MiB free

llama-server stderr: load: 0 unused tokens

llama-server stderr: load: printing all EOG tokens:
load:   - 248044 ('<|endoftext|>')
load:   - 248046 ('<|im_end|>')
load:   - 248063 ('<|fim_pad|>')
load:   - 248064 ('<|repo_name|>')
load:   - 248065 ('<|file_sep|>')

llama-server stderr: load: special tokens cache size = 33

llama-server stderr: load: token to piece cache size = 1.7581 MB
print_info: arch                  = qwen35moe
print_info: vocab_only            = 0
print_info: no_alloc              = 0
print_info: n_ctx_train           = 262144
print_info: n_embd                = 2048
print_info: n_embd_inp            = 2048
print_info: n_layer               = 41
print_info: n_head                = 16
print_info: n_head_kv             = 2
print_info: n_rot                 = 64
print_info: n_swa                 = 0
print_info: is_swa_any            = 0
print_info: n_embd_head_k         = 256
print_info: n_embd_head_v         = 256
print_info: n_gqa                 = 8
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: f_attn_value_scale    = 0.0000
print_info: n_ff                  = 0
print_info: n_expert              = 256
print_info: n_expert_used         = 8
print_info: n_expert_groups       = 0
print_info: n_group_used          = 0
print_info: causal attn           = 1
print_info: pooling type          = -1
print_info: rope type             = 40
print_info: rope scaling          = linear
print_info: freq_base_train       = 10000000.0
print_info: freq_scale_train      = 1
print_info: n_ctx_orig_yarn       = 262144
print_info: rope_yarn_log_mul     = 0.0000
print_info: rope_finetuned        = unknown
print_info: mrope sections        = [11, 11, 10, 0]
print_info: ssm_d_conv            = 4
print_info: ssm_d_inner           = 4096
print_info: ssm_d_state           = 128
print_info: ssm_dt_rank           = 32
print_info: ssm_n_group           = 16
print_info: ssm_dt_b_c_rms        = 0
print_info: model type            = ?B
print_info: model params          = 35.51 B
print_info: general.name          = Qwen3.6-35B-A3B
print_info: vocab type            = BPE
print_info: n_vocab               = 248320
print_info: n_merges              = 247587
print_info: BOS token             = 248044 '<|endoftext|>'
print_info: EOS token             = 248046 '<|im_end|>'
print_info: EOT token             = 248046 '<|im_end|>'
print_info: PAD token             = 248055 '<|vision_pad|>'
print_info: LF token              = 198 'Ċ'
print_info: FIM PRE token         = 248060 '<|fim_prefix|>'
print_info: FIM SUF token         = 248062 '<|fim_suffix|>'
print_info: FIM MID token         = 248061 '<|fim_middle|>'
print_info: FIM PAD token         = 248063 '<|fim_pad|>'
print_info: FIM REP token         = 248064 '<|repo_name|>'
print_info: FIM SEP token         = 248065 '<|file_sep|>'
print_info: EOG token             = 248044 '<|endoftext|>'
print_info: EOG token             = 248046 '<|im_end|>'
print_info: EOG token             = 248063 '<|fim_pad|>'
print_info: EOG token             = 248064 '<|repo_name|>'
print_info: EOG token             = 248065 '<|file_sep|>'
print_info: max token length      = 256
load_tensors: loading model tensors, this can take a while... (mmap = true, direct_io = false)

llama-server stderr: llama_model_load: error loading model: missing tensor 'blk.40.ssm_conv1d.weight'
llama_model_load_from_file_impl: failed to load model
common_init_from_params: failed to load model 'E:\llm-model\unsloth\Qwen3.6-35B-A3B-MTP-GGUF\Qwen3.6-35B-A3B-UD-IQ4_NL.gguf'
srv    load_model: failed to load model, 'E:\llm-model\unsloth\Qwen3.6-35B-A3B-MTP-GGUF\Qwen3.6-35B-A3B-UD-IQ4_NL.gguf'
srv    operator(): operator(): cleaning up before exit...

llama-server stderr: main: exiting due to model loading error

[ServeurpersoCom]

Normal, not SSM but MTP. Work in progress here: #22673

[QiWu666]

PS    .\llama-server.exe `
>>   -m ".\Qwen3.6-35B-A3B-Uncensored-Genesis-MTP-APEX-Compact.gguf" `
>>   --jinja `
>>   -ngl 99 `
>>   -c 262144 `
>>   -fa on `
>>   -t 8 `
>>   -b 4096 `
>>   --cache-type-k q4_1 `
>>   --cache-type-v q4_1 `
>>   --host 0.0.0.0 `
>>   --port 8080 `
>>   --no-mmap `
>>   -np 1 `
>>   --temp 0.6 `
>>   --top-k 40 `
>>   --top-p 0.95
ggml_cuda_init: found 1 CUDA devices (Total VRAM: 22527 MiB):
  Device 0: NVIDIA GeForce RTX 2080 Ti, compute capability 7.5, VMM: yes, VRAM: 22527 MiB
build_info: b1-4d24ad8
system_info: n_threads = 8 (n_threads_batch = 8) / 12 | CUDA : ARCHS = 500,610,700,750,800,860,890 | USE_GRAPHS = 1 | PEER_MAX_BATCH_SIZE = 128 | FA_ALL_QUANTS = 1 | CPU : SSE3 = 1 | SSSE3 = 1 | AVX = 1 | AVX2 = 1 | F16C = 1 | FMA = 1 | BMI2 = 1 | LLAMAFILE = 1 | OPENMP = 1 | REPACK = 1 |
Running without SSL
init: using 11 threads for HTTP server
start: binding port with default address family
main: loading model
srv    load_model: loading model '.\Qwen3.6-35B-A3B-Uncensored-Genesis-MTP-APEX-Compact.gguf'
common_init_result: fitting params to device memory, for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on
llama_model_load: error loading model: missing tensor 'blk.40.ssm_conv1d.weight'
llama_model_load_from_file_impl: failed to load model
llama_params_fit: encountered an error while trying to fit params to free device memory: failed to load model
llama_params_fit: fitting params to free memory took 0.75 seconds
llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GeForce RTX 2080 Ti) (0000:26:00.0) - 21326 MiB free
llama_model_loader: loaded meta data with 46 key-value pairs and 753 tensors from .\Qwen3.6-35B-A3B-Uncensored-Genesis-MTP-APEX-Compact.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              = qwen35moe
llama_model_loader: - kv   1:                               general.type str              = model
llama_model_loader: - kv   2:                     general.sampling.top_k i32              = 20
llama_model_loader: - kv   3:                     general.sampling.top_p f32              = 0.950000
llama_model_loader: - kv   4:                      general.sampling.temp f32              = 1.000000
llama_model_loader: - kv   5:                               general.name str              = Qwen3.6-35B-A3B-Uncensored-HauhauCS-A...
llama_model_loader: - kv   6:                           general.finetune str              = 3Ref
llama_model_loader: - kv   7:                           general.basename str              = KL0.0764
llama_model_loader: - kv   8:                   qwen35moe.context_length u32              = 262144
llama_model_loader: - kv   9:                 qwen35moe.embedding_length u32              = 2048
llama_model_loader: - kv  10:             qwen35moe.attention.head_count u32              = 16
llama_model_loader: - kv  11:          qwen35moe.attention.head_count_kv u32              = 2
llama_model_loader: - kv  12:          qwen35moe.rope.dimension_sections arr[i32,4]       = [11, 11, 10, 0]
llama_model_loader: - kv  13:                   qwen35moe.rope.freq_base f32              = 10000000.000000
llama_model_loader: - kv  14: qwen35moe.attention.layer_norm_rms_epsilon f32              = 0.000001
llama_model_loader: - kv  15:                     qwen35moe.expert_count u32              = 256
llama_model_loader: - kv  16:                qwen35moe.expert_used_count u32              = 8
llama_model_loader: - kv  17:             qwen35moe.attention.key_length u32              = 256
llama_model_loader: - kv  18:           qwen35moe.attention.value_length u32              = 256
llama_model_loader: - kv  19:       qwen35moe.expert_feed_forward_length u32              = 512
llama_model_loader: - kv  20: qwen35moe.expert_shared_feed_forward_length u32              = 512
llama_model_loader: - kv  21:                  qwen35moe.ssm.conv_kernel u32              = 4
llama_model_loader: - kv  22:                   qwen35moe.ssm.state_size u32              = 128
llama_model_loader: - kv  23:                  qwen35moe.ssm.group_count u32              = 16
llama_model_loader: - kv  24:               qwen35moe.ssm.time_step_rank u32              = 32
llama_model_loader: - kv  25:                   qwen35moe.ssm.inner_size u32              = 4096
llama_model_loader: - kv  26:          qwen35moe.full_attention_interval u32              = 4
llama_model_loader: - kv  27:             qwen35moe.rope.dimension_count u32              = 64
llama_model_loader: - kv  28:                       tokenizer.ggml.model str              = gpt2
llama_model_loader: - kv  29:                         tokenizer.ggml.pre str              = qwen35
llama_model_loader: - kv  30:                      tokenizer.ggml.tokens arr[str,248320]  = ["!", "\"", "#", "$", "%", "&", "'", ...
llama_model_loader: - kv  31:                  tokenizer.ggml.token_type arr[i32,248320]  = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv  32:                      tokenizer.ggml.merges arr[str,247587]  = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv  33:                tokenizer.ggml.eos_token_id u32              = 248046
llama_model_loader: - kv  34:            tokenizer.ggml.padding_token_id u32              = 248044
llama_model_loader: - kv  35:                tokenizer.ggml.bos_token_id u32              = 248044
llama_model_loader: - kv  36:                    tokenizer.chat_template str              = {%- set image_count = namespace(value...
llama_model_loader: - kv  37:                         general.size_label str              = 35B-A3B
llama_model_loader: - kv  38:                      quantize.imatrix.file str              = Qwen3.6-35B-A3B-Uncensored-HauhauCS-A...
llama_model_loader: - kv  39:                   quantize.imatrix.dataset str              = /home/hauhaucs/llama.cpp/groups_merge...
llama_model_loader: - kv  40:             quantize.imatrix.entries_count u32              = 510
llama_model_loader: - kv  41:              quantize.imatrix.chunks_count u32              = 93
llama_model_loader: - kv  42:                      qwen35moe.block_count u32              = 41
llama_model_loader: - kv  43:             qwen35moe.nextn_predict_layers u32              = 1
llama_model_loader: - kv  44:               general.quantization_version u32              = 2
llama_model_loader: - kv  45:                          general.file_type u32              = 15
llama_model_loader: - type  f32:  308 tensors
llama_model_loader: - type q8_0:   11 tensors
llama_model_loader: - type q3_K:   93 tensors
llama_model_loader: - type q4_K:  173 tensors
llama_model_loader: - type q5_K:   30 tensors
llama_model_loader: - type q6_K:  136 tensors
llama_model_loader: - type bf16:    2 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type   = Q4_K - Medium
print_info: file size   = 16.77 GiB (4.06 BPW)
load: 0 unused tokens
load: printing all EOG tokens:
load:   - 248044 ('<|endoftext|>')
load:   - 248046 ('<|im_end|>')
load:   - 248063 ('<|fim_pad|>')
load:   - 248064 ('<|repo_name|>')
load:   - 248065 ('<|file_sep|>')
load: special tokens cache size = 33
load: token to piece cache size = 1.7581 MB
print_info: arch                  = qwen35moe
print_info: vocab_only            = 0
print_info: no_alloc              = 0
print_info: n_ctx_train           = 262144
print_info: n_embd                = 2048
print_info: n_embd_inp            = 2048
print_info: n_layer               = 41
print_info: n_head                = 16
print_info: n_head_kv             = 2
print_info: n_rot                 = 64
print_info: n_swa                 = 0
print_info: is_swa_any            = 0
print_info: n_embd_head_k         = 256
print_info: n_embd_head_v         = 256
print_info: n_gqa                 = 8
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                  = 0
print_info: n_expert              = 256
print_info: n_expert_used         = 8
print_info: n_expert_groups       = 0
print_info: n_group_used          = 0
print_info: causal attn           = 1
print_info: pooling type          = -1
print_info: rope type             = 40
print_info: rope scaling          = linear
print_info: freq_base_train       = 10000000.0
print_info: freq_scale_train      = 1
print_info: n_ctx_orig_yarn       = 262144
print_info: rope_yarn_log_mul     = 0.0000
print_info: rope_finetuned        = unknown
print_info: mrope sections        = [11, 11, 10, 0]
print_info: ssm_d_conv            = 4
print_info: ssm_d_inner           = 4096
print_info: ssm_d_state           = 128
print_info: ssm_dt_rank           = 32
print_info: ssm_n_group           = 16
print_info: ssm_dt_b_c_rms        = 0
print_info: model type            = ?B
print_info: model params          = 35.51 B
print_info: general.name          = Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive
print_info: vocab type            = BPE
print_info: n_vocab               = 248320
print_info: n_merges              = 247587
print_info: BOS token             = 248044 '<|endoftext|>'
print_info: EOS token             = 248046 '<|im_end|>'
print_info: EOT token             = 248046 '<|im_end|>'
print_info: PAD token             = 248044 '<|endoftext|>'
print_info: LF token              = 198 'Ċ'
print_info: FIM PRE token         = 248060 '<|fim_prefix|>'
print_info: FIM SUF token         = 248062 '<|fim_suffix|>'
print_info: FIM MID token         = 248061 '<|fim_middle|>'
print_info: FIM PAD token         = 248063 '<|fim_pad|>'
print_info: FIM REP token         = 248064 '<|repo_name|>'
print_info: FIM SEP token         = 248065 '<|file_sep|>'
print_info: EOG token             = 248044 '<|endoftext|>'
print_info: EOG token             = 248046 '<|im_end|>'
print_info: EOG token             = 248063 '<|fim_pad|>'
print_info: EOG token             = 248064 '<|repo_name|>'
print_info: EOG token             = 248065 '<|file_sep|>'
print_info: max token length      = 256
load_tensors: loading model tensors, this can take a while... (mmap = false, direct_io = false)
llama_model_load: error loading model: missing tensor 'blk.40.ssm_conv1d.weight'
llama_model_load_from_file_impl: failed to load model
common_init_from_params: failed to load model '.\Qwen3.6-35B-A3B-Uncensored-Genesis-MTP-APEX-Compact.gguf'
srv    load_model: failed to load model, '.\Qwen3.6-35B-A3B-Uncensored-Genesis-MTP-APEX-Compact.gguf'
srv   operator (): operator (): cleaning up before exit...
main: exiting due to model loading error

same question

[kamzil]

Normal, not SSM but MTP. Work in progress here: #22673

Still getting this error on Strix Halo, even though that PR was merged. Model: unsloth/Qwen3.6-35B-A3B-MTP-GGUF:UD-Q8_K_XL

[batot1]

On Linux I have this same:

-m /models/qwen36/Qwen_Qwen3.6-35B-A3B-Q8_0.gguf --host 0.0.0.0 --port 8081 --ctx-size 262144 --parallel 1 -b 8192 -ub 8192 -fa on --no-warmup --cache-type-k q8_0 --cache-type-v q8_0 --threads 4
ggml_cuda_init: found 1 ROCm devices (Total VRAM: 32752 MiB):
Device 0: , gfx906:sramecc+:xnack- (0x906), VMM: no, Wave Size: 64, VRAM: 32752 MiB
build: 32 (c9790f9) with GNU 11.4.0 for Linux x86_64
system info: n_threads = 4, n_threads_batch = 4, total_threads = 12

system_info: n_threads = 4 (n_threads_batch = 4) / 12 | ROCm : NO_VMM = 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 |

init: using 11 threads for HTTP server
start: binding port with default address family
main: loading model
srv load_model: loading model '/models/qwen36/Qwen_Qwen3.6-35B-A3B-Q8_0.gguf'
common_init_result: fitting params to device memory, for bugs during this step try to reproduce them with -fit off, or provide --verbose logs if the bug only occurs with -fit on
llama_model_load: error loading model: missing tensor 'blk.40.ssm_conv1d.weight'
llama_model_load_from_file_impl: failed to load model
llama_params_fit: encountered an error while trying to fit params to free device memory: failed to load model
llama_params_fit: fitting params to free memory took 0.52 seconds
llama_model_load_from_file_impl: using device ROCm0 () (0000:09:00.0) - 32732 MiB free
llama_model_loader: loaded meta data with 49 key-value pairs and 753 tensors from /models/qwen36/Qwen_Qwen3.6-35B-A3B-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 = qwen35moe
llama_model_loader: - kv 1: general.type str = model
llama_model_loader: - kv 2: general.sampling.top_k i32 = 20
llama_model_loader: - kv 3: general.sampling.top_p f32 = 0.950000
llama_model_loader: - kv 4: general.sampling.temp f32 = 1.000000
llama_model_loader: - kv 5: general.name str = Qwen3.6 35B A3B
llama_model_loader: - kv 6: general.basename str = Qwen3.6
llama_model_loader: - kv 7: general.size_label str = 35B-A3B
llama_model_loader: - kv 8: general.license str = apache-2.0
llama_model_loader: - kv 9: general.license.link str = https://huggingface.co/Qwen/Qwen3.6-3...
llama_model_loader: - kv 10: general.tags arr[str,1] = ["image-text-to-text"]
llama_model_loader: - kv 11: qwen35moe.block_count u32 = 41
llama_model_loader: - kv 12: qwen35moe.context_length u32 = 262144
llama_model_loader: - kv 13: qwen35moe.embedding_length u32 = 2048
llama_model_loader: - kv 14: qwen35moe.attention.head_count u32 = 16
llama_model_loader: - kv 15: qwen35moe.attention.head_count_kv u32 = 2
llama_model_loader: - kv 16: qwen35moe.rope.dimension_sections arr[i32,4] = [11, 11, 10, 0]
llama_model_loader: - kv 17: qwen35moe.rope.freq_base f32 = 10000000.000000
llama_model_loader: - kv 18: qwen35moe.attention.layer_norm_rms_epsilon f32 = 0.000001
llama_model_loader: - kv 19: qwen35moe.expert_count u32 = 256
llama_model_loader: - kv 20: qwen35moe.expert_used_count u32 = 8
llama_model_loader: - kv 21: qwen35moe.attention.key_length u32 = 256
llama_model_loader: - kv 22: qwen35moe.attention.value_length u32 = 256
llama_model_loader: - kv 23: qwen35moe.expert_feed_forward_length u32 = 512
llama_model_loader: - kv 24: qwen35moe.expert_shared_feed_forward_length u32 = 512
llama_model_loader: - kv 25: qwen35moe.ssm.conv_kernel u32 = 4
llama_model_loader: - kv 26: qwen35moe.ssm.state_size u32 = 128
llama_model_loader: - kv 27: qwen35moe.ssm.group_count u32 = 16
llama_model_loader: - kv 28: qwen35moe.ssm.time_step_rank u32 = 32
llama_model_loader: - kv 29: qwen35moe.ssm.inner_size u32 = 4096
llama_model_loader: - kv 30: qwen35moe.full_attention_interval u32 = 4
llama_model_loader: - kv 31: qwen35moe.rope.dimension_count u32 = 64
llama_model_loader: - kv 32: qwen35moe.nextn_predict_layers u32 = 1
llama_model_loader: - kv 33: tokenizer.ggml.model str = gpt2
llama_model_loader: - kv 34: tokenizer.ggml.pre str = qwen35
llama_model_loader: - kv 35: tokenizer.ggml.tokens arr[str,248320] = ["!", """, "#", "$", "%", "&", "'", ...
llama_model_loader: - kv 36: tokenizer.ggml.token_type arr[i32,248320] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
llama_model_loader: - kv 37: tokenizer.ggml.merges arr[str,247587] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",...
llama_model_loader: - kv 38: tokenizer.ggml.eos_token_id u32 = 248046
llama_model_loader: - kv 39: tokenizer.ggml.padding_token_id u32 = 248044
llama_model_loader: - kv 40: tokenizer.ggml.bos_token_id u32 = 248044
llama_model_loader: - kv 41: tokenizer.ggml.add_bos_token bool = false
llama_model_loader: - kv 42: tokenizer.chat_template str = {%- set image_count = namespace(value...
llama_model_loader: - kv 43: general.quantization_version u32 = 2
llama_model_loader: - kv 44: general.file_type u32 = 7
llama_model_loader: - kv 45: quantize.imatrix.file str = /models_out/Qwen3.6-35B-A3B-GGUF/Qwen...
llama_model_loader: - kv 46: quantize.imatrix.dataset str = /training_dir/calibration_datav5.txt
llama_model_loader: - kv 47: quantize.imatrix.entries_count u32 = 510
llama_model_loader: - kv 48: quantize.imatrix.chunks_count u32 = 802
llama_model_loader: - type f32: 368 tensors
llama_model_loader: - type q8_0: 383 tensors
llama_model_loader: - type bf16: 2 tensors
print_info: file format = GGUF V3 (latest)
print_info: file type = Q8_0
print_info: file size = 35.21 GiB (8.52 BPW)
load: 0 unused tokens
load: printing all EOG tokens:
load: - 248044 ('<|endoftext|>')
load: - 248046 ('<|im_end|>')
load: - 248063 ('<|fim_pad|>')
load: - 248064 ('<|repo_name|>')
load: - 248065 ('<|file_sep|>')
load: special tokens cache size = 33
load: token to piece cache size = 1.7581 MB
print_info: arch = qwen35moe
print_info: vocab_only = 0
print_info: no_alloc = 0
print_info: n_ctx_train = 262144
print_info: n_embd = 2048
print_info: n_embd_inp = 2048
print_info: n_layer = 41
print_info: n_head = 16
print_info: n_head_kv = 2
print_info: n_rot = 64
print_info: n_swa = 0
print_info: is_swa_any = 0
print_info: n_embd_head_k = 256
print_info: n_embd_head_v = 256
print_info: n_gqa = 8
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 = 0
print_info: n_expert = 256
print_info: n_expert_used = 8
print_info: n_expert_groups = 0
print_info: n_group_used = 0
print_info: causal attn = 1
print_info: pooling type = -1
print_info: rope type = 40
print_info: rope scaling = linear
print_info: freq_base_train = 10000000.0
print_info: freq_scale_train = 1
print_info: n_ctx_orig_yarn = 262144
print_info: rope_yarn_log_mul = 0.0000
print_info: rope_finetuned = unknown
print_info: mrope sections = [11, 11, 10, 0]
print_info: ssm_d_conv = 4
print_info: ssm_d_inner = 4096
print_info: ssm_d_state = 128
print_info: ssm_dt_rank = 32
print_info: ssm_n_group = 16
print_info: ssm_dt_b_c_rms = 0
print_info: model type = ?B
print_info: model params = 35.51 B
print_info: general.name = Qwen3.6 35B A3B
print_info: vocab type = BPE
print_info: n_vocab = 248320
print_info: n_merges = 247587
print_info: BOS token = 248044 '<|endoftext|>'
print_info: EOS token = 248046 '<|im_end|>'
print_info: EOT token = 248046 '<|im_end|>'
print_info: PAD token = 248044 '<|endoftext|>'
print_info: LF token = 198 'Ċ'
print_info: FIM PRE token = 248060 '<|fim_prefix|>'
print_info: FIM SUF token = 248062 '<|fim_suffix|>'
print_info: FIM MID token = 248061 '<|fim_middle|>'
print_info: FIM PAD token = 248063 '<|fim_pad|>'
print_info: FIM REP token = 248064 '<|repo_name|>'
print_info: FIM SEP token = 248065 '<|file_sep|>'
print_info: EOG token = 248044 '<|endoftext|>'
print_info: EOG token = 248046 '<|im_end|>'
print_info: EOG token = 248063 '<|fim_pad|>'
print_info: EOG token = 248064 '<|repo_name|>'
print_info: EOG token = 248065 '<|file_sep|>'
print_info: max token length = 256
load_tensors: loading model tensors, this can take a while... (mmap = true, direct_io = false)
llama_model_load: error loading model: missing tensor 'blk.40.ssm_conv1d.weight'
llama_model_load_from_file_impl: failed to load model
common_init_from_params: failed to load model '/models/qwen36/Qwen_Qwen3.6-35B-A3B-Q8_0.gguf'
srv load_model: failed to load model, '/models/qwen36/Qwen_Qwen3.6-35B-A3B-Q8_0.gguf'
srv operator(): operator(): cleaning up before exit...
main: exiting due to model loading error