doesn't work in `lmsysorg/sglang:dev-cu13` on nvidia dgx spark
#1
by Slach - opened
use following docker compose
serivces:
sglang:
image: lmsysorg/sglang:dev-cu13
volumes:
- ${HOME}/.cache/huggingface:/models/huggingface:ro
shm_size: '32gb'
networks:
- spark-network
environment:
- HF_HOME=/models/huggingface
- HF_HUB_OFFLINE=1
- HF_TOKEN=${HF_TOKEN:-}
ports:
- "127.0.0.1:30000:30000"
command: |
python3 -m sglang.launch_server
--port 30000
--host 0.0.0.0
--model-path AxionML--Qwen3.5-35B-A3B
--context-length 261375
--kv-cache-dtype fp8_e4m3
--tp 1
--reasoning-parser qwen3
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: 1
capabilities: [ gpu ]
got error during loading
sglang-1 | [2026-03-20 03:14:18] server_args=ServerArgs(model_path='AxionML/Qwen3.5-35B-A3B-NVFP4', tokenizer_path='AxionML/Qwen3.5-35B-A3B-NVFP4', tokenizer_mode='auto', tokenizer_worker_num=1, skip_tokenizer_init=False, load_format='auto', model_loader_extra_config='{}', trust_remote_code=False, context_length=261375, is_embedding=False, enable_multimodal=None, revision=None, model_impl='auto', host='0.0.0.0', port=30000, fastapi_root_path='', grpc_mode=False, skip_server_warmup=False, warmups=None, nccl_port=None, checkpoint_engine_wait_weights_before_ready=False, ssl_keyfile=None, ssl_certfile=None, ssl_ca_certs=None, ssl_keyfile_password=None, enable_ssl_refresh=False, dtype='auto', quantization=None, quantization_param_path=None, kv_cache_dtype='fp8_e4m3', enable_fp32_lm_head=False, modelopt_quant=None, modelopt_checkpoint_restore_path=None, modelopt_checkpoint_save_path=None, modelopt_export_path=None, quantize_and_serve=False, rl_quant_profile=None, mem_fraction_static=0.8248909375, max_running_requests=None, max_queued_requests=None, max_total_tokens=None, chunked_prefill_size=8192, enable_dynamic_chunking=False, max_prefill_tokens=16384, prefill_max_requests=None, schedule_policy='fcfs', enable_priority_scheduling=False, disable_priority_preemption=False, default_priority_value=None, abort_on_priority_when_disabled=False, schedule_low_priority_values_first=False, priority_scheduling_preemption_threshold=10, schedule_conservativeness=1.0, page_size=1, swa_full_tokens_ratio=0.8, disable_hybrid_swa_memory=False, radix_eviction_policy='lru', enable_prefill_delayer=False, prefill_delayer_max_delay_passes=30, prefill_delayer_token_usage_low_watermark=None, prefill_delayer_forward_passes_buckets=None, prefill_delayer_wait_seconds_buckets=None, device='cuda', tp_size=1, pp_size=1, pp_max_micro_batch_size=None, pp_async_batch_depth=0, stream_interval=1, incremental_streaming_output=False, enable_streaming_session=False, random_seed=525567074, constrained_json_whitespace_pattern=None, constrained_json_disable_any_whitespace=False, watchdog_timeout=300, soft_watchdog_timeout=None, dist_timeout=None, download_dir=None, model_checksum=None, base_gpu_id=0, gpu_id_step=1, sleep_on_idle=False, use_ray=False, custom_sigquit_handler=None, log_level='info', log_level_http=None, log_requests=False, log_requests_level=2, log_requests_format='text', log_requests_target=None, uvicorn_access_log_exclude_prefixes=[], crash_dump_folder=None, show_time_cost=False, enable_metrics=False, enable_metrics_for_all_schedulers=False, tokenizer_metrics_custom_labels_header='x-custom-labels', tokenizer_metrics_allowed_custom_labels=None, extra_metric_labels=None, bucket_time_to_first_token=None, bucket_inter_token_latency=None, bucket_e2e_request_latency=None, collect_tokens_histogram=False, prompt_tokens_buckets=None, generation_tokens_buckets=None, gc_warning_threshold_secs=0.0, decode_log_interval=40, enable_request_time_stats_logging=False, kv_events_config=None, enable_trace=False, otlp_traces_endpoint='localhost:4317', export_metrics_to_file=False, export_metrics_to_file_dir=None, api_key=None, admin_api_key=None, served_model_name='AxionML/Qwen3.5-35B-A3B-NVFP4', weight_version='default', chat_template=None, hf_chat_template_name=None, completion_template=None, file_storage_path='sglang_storage', enable_cache_report=False, reasoning_parser='qwen3', tool_call_parser=None, tool_server=None, sampling_defaults='model', dp_size=1, load_balance_method='round_robin', attn_cp_size=1, moe_dp_size=1, dist_init_addr=None, nnodes=1, node_rank=0, json_model_override_args='{}', preferred_sampling_params=None, enable_lora=None, enable_lora_overlap_loading=None, max_lora_rank=None, lora_target_modules=None, lora_paths=None, max_loaded_loras=None, max_loras_per_batch=8, lora_eviction_policy='lru', lora_backend='csgmv', max_lora_chunk_size=16, attention_backend='flashinfer', decode_attention_backend=None, prefill_attention_backend=None, sampling_backend='flashinfer', grammar_backend='xgrammar', mm_attention_backend=None, fp8_gemm_runner_backend='auto', fp4_gemm_runner_backend='auto', nsa_prefill_backend=None, nsa_decode_backend=None, disable_flashinfer_autotune=False, mamba_backend='triton', speculative_algorithm=None, speculative_draft_model_path=None, speculative_draft_model_revision=None, speculative_draft_load_format=None, speculative_num_steps=None, speculative_eagle_topk=None, speculative_num_draft_tokens=None, speculative_accept_threshold_single=1.0, speculative_accept_threshold_acc=1.0, speculative_token_map=None, speculative_attention_mode='prefill', speculative_draft_attention_backend=None, speculative_moe_runner_backend='auto', speculative_moe_a2a_backend=None, speculative_draft_model_quantization=None, speculative_ngram_min_match_window_size=1, speculative_ngram_max_match_window_size=12, speculative_ngram_min_bfs_breadth=1, speculative_ngram_max_bfs_breadth=10, speculative_ngram_match_type='BFS', speculative_ngram_branch_length=18, speculative_ngram_capacity=10000000, enable_multi_layer_eagle=False, ep_size=1, moe_a2a_backend='none', moe_runner_backend='auto', flashinfer_mxfp4_moe_precision='default', enable_flashinfer_allreduce_fusion=False, enable_aiter_allreduce_fusion=False, deepep_mode='auto', ep_num_redundant_experts=0, ep_dispatch_algorithm=None, init_expert_location='trivial', enable_eplb=False, eplb_algorithm='auto', eplb_rebalance_num_iterations=1000, eplb_rebalance_layers_per_chunk=None, eplb_min_rebalancing_utilization_threshold=1.0, expert_distribution_recorder_mode=None, expert_distribution_recorder_buffer_size=1000, enable_expert_distribution_metrics=False, deepep_config=None, moe_dense_tp_size=None, elastic_ep_backend=None, enable_elastic_expert_backup=False, mooncake_ib_device=None, max_mamba_cache_size=None, mamba_ssm_dtype=None, mamba_full_memory_ratio=0.9, mamba_scheduler_strategy='no_buffer', mamba_track_interval=256, linear_attn_backend='triton', linear_attn_decode_backend=None, linear_attn_prefill_backend=None, enable_hierarchical_cache=False, hicache_ratio=2.0, hicache_size=0, hicache_write_policy='write_through', hicache_io_backend='kernel', hicache_mem_layout='layer_first', disable_hicache_numa_detect=False, hicache_storage_backend=None, hicache_storage_prefetch_policy='best_effort', hicache_storage_backend_extra_config=None, hierarchical_sparse_attention_extra_config=None, enable_lmcache=False, kt_weight_path=None, kt_method='AMXINT4', kt_cpuinfer=None, kt_threadpool_count=2, kt_num_gpu_experts=None, kt_max_deferred_experts_per_token=None, dllm_algorithm=None, dllm_algorithm_config=None, enable_double_sparsity=False, ds_channel_config_path=None, ds_heavy_channel_num=32, ds_heavy_token_num=256, ds_heavy_channel_type='qk', ds_sparse_decode_threshold=4096, cpu_offload_gb=0, offload_group_size=-1, offload_num_in_group=1, offload_prefetch_step=1, offload_mode='cpu', multi_item_scoring_delimiter=None, disable_radix_cache=False, cuda_graph_max_bs=256, cuda_graph_bs=[1, 2, 4, 8, 12, 16, 24, 32, 40, 48, 56, 64, 72, 80, 88, 96, 104, 112, 120, 128, 136, 144, 152, 160, 168, 176, 184, 192, 200, 208, 216, 224, 232, 240, 248, 256], disable_cuda_graph=False, disable_cuda_graph_padding=False, enable_profile_cuda_graph=False, enable_cudagraph_gc=False, enable_layerwise_nvtx_marker=False, enable_nccl_nvls=False, enable_symm_mem=False, disable_flashinfer_cutlass_moe_fp4_allgather=False, enable_tokenizer_batch_encode=False, disable_tokenizer_batch_decode=False, disable_outlines_disk_cache=False, disable_custom_all_reduce=False, enable_mscclpp=False, enable_torch_symm_mem=False, pre_warm_nccl=False, disable_overlap_schedule=True, enable_mixed_chunk=False, enable_dp_attention=False, enable_dp_lm_head=False, enable_two_batch_overlap=False, enable_single_batch_overlap=False, tbo_token_distribution_threshold=0.48, enable_torch_compile=False, disable_piecewise_cuda_graph=True, enforce_piecewise_cuda_graph=False, enable_torch_compile_debug_mode=False, torch_compile_max_bs=32, piecewise_cuda_graph_max_tokens=8192, piecewise_cuda_graph_tokens=[4, 8, 12, 16, 20, 24, 28, 32, 48, 64, 80, 96, 112, 128, 144, 160, 176, 192, 208, 224, 240, 256, 288, 320, 352, 384, 416, 448, 480, 512, 576, 640, 704, 768, 832, 896, 960, 1024, 1280, 1536, 1792, 2048, 2304, 2560, 2816, 3072, 3328, 3584, 3840, 4096, 4608, 5120, 5632, 6144, 6656, 7168, 7680, 8192], piecewise_cuda_graph_compiler='eager', torchao_config='', enable_nan_detection=False, enable_p2p_check=False, triton_attention_reduce_in_fp32=False, triton_attention_num_kv_splits=8, triton_attention_split_tile_size=None, num_continuous_decode_steps=1, delete_ckpt_after_loading=False, enable_memory_saver=False, enable_weights_cpu_backup=False, enable_draft_weights_cpu_backup=False, allow_auto_truncate=False, enable_custom_logit_processor=False, flashinfer_mla_disable_ragged=False, disable_shared_experts_fusion=False, disable_chunked_prefix_cache=False, disable_fast_image_processor=False, keep_mm_feature_on_device=False, enable_return_hidden_states=False, enable_return_routed_experts=False, scheduler_recv_interval=1, numa_node=None, enable_deterministic_inference=False, rl_on_policy_target=None, enable_attn_tp_input_scattered=False, enable_nsa_prefill_context_parallel=False, nsa_prefill_cp_mode='round-robin-split', enable_fused_qk_norm_rope=False, enable_precise_embedding_interpolation=False, enable_fused_moe_sum_all_reduce=False, enable_dynamic_batch_tokenizer=False, dynamic_batch_tokenizer_batch_size=32, dynamic_batch_tokenizer_batch_timeout=0.002, debug_tensor_dump_output_folder=None, debug_tensor_dump_layers=None, debug_tensor_dump_input_file=None, debug_tensor_dump_inject=False, disaggregation_mode='null', disaggregation_transfer_backend='mooncake', disaggregation_bootstrap_port=8998, disaggregation_ib_device=None, disaggregation_decode_enable_offload_kvcache=False, num_reserved_decode_tokens=512, disaggregation_decode_polling_interval=1, encoder_only=False, language_only=False, encoder_transfer_backend='zmq_to_scheduler', encoder_urls=[], enable_adaptive_dispatch_to_encoder=False, custom_weight_loader=[], weight_loader_disable_mmap=False, remote_instance_weight_loader_seed_instance_ip=None, remote_instance_weight_loader_seed_instance_service_port=None, remote_instance_weight_loader_send_weights_group_ports=None, remote_instance_weight_loader_backend='nccl', remote_instance_weight_loader_start_seed_via_transfer_engine=False, modelexpress_config=None, enable_pdmux=False, pdmux_config_path=None, sm_group_num=8, mm_max_concurrent_calls=32, mm_per_request_timeout=10.0, enable_broadcast_mm_inputs_process=False, enable_prefix_mm_cache=False, mm_enable_dp_encoder=False, mm_process_config={}, limit_mm_data_per_request=None, enable_mm_global_cache=False, decrypted_config_file=None, decrypted_draft_config_file=None, forward_hooks=None)
sglang-1 | [2026-03-20 03:14:18] DeepGemm is enabled but the scale_fmt of checkpoint is not ue8m0. This might cause accuracy degradation on Blackwell.
sglang-1 | [2026-03-20 03:14:18] Transformers version 5.3.0 is used for model type qwen3_5_moe. If you experience issues related to RoPE parameters, they may be due to incompatibilities between Transformers >=5.0.0 and some models. You can try downgrading to transformers==4.57.1 as a workaround.
sglang-1 | /usr/local/lib/python3.12/dist-packages/torch/cuda/__init__.py:283: UserWarning:
sglang-1 | Found GPU0 NVIDIA GB10 which is of cuda capability 12.1.
sglang-1 | Minimum and Maximum cuda capability supported by this version of PyTorch is
sglang-1 | (8.0) - (12.0)
sglang-1 |
sglang-1 | warnings.warn(
sglang-1 | /usr/local/lib/python3.12/dist-packages/torch/cuda/__init__.py:283: UserWarning:
sglang-1 | Found GPU0 NVIDIA GB10 which is of cuda capability 12.1.
sglang-1 | Minimum and Maximum cuda capability supported by this version of PyTorch is
sglang-1 | (8.0) - (12.0)
sglang-1 |
sglang-1 | warnings.warn(
sglang-1 | [2026-03-20 03:14:21] Using default HuggingFace chat template with detected content format: openai
sglang-1 | [2026-03-20 03:14:22] DeepGemm is enabled but the scale_fmt of checkpoint is not ue8m0. This might cause accuracy degradation on Blackwell.
sglang-1 | [2026-03-20 03:14:22] Transformers version 5.3.0 is used for model type qwen3_5_moe. If you experience issues related to RoPE parameters, they may be due to incompatibilities between Transformers >=5.0.0 and some models. You can try downgrading to transformers==4.57.1 as a workaround.
sglang-1 | [2026-03-20 03:14:23] Detected generic TokenizersBackend for AxionML/Qwen3.5-35B-A3B-NVFP4, retrying with trust_remote_code=True
sglang-1 | [2026-03-20 03:14:24] SM120 (Blackwell) detected: auto-selecting fp4-gemm-backend=flashinfer_cudnn
sglang-1 | [2026-03-20 03:14:24] DeepGemm is enabled but the scale_fmt of checkpoint is not ue8m0. This might cause accuracy degradation on Blackwell.
sglang-1 | [2026-03-20 03:14:24] Transformers version 5.3.0 is used for model type qwen3_5_moe. If you experience issues related to RoPE parameters, they may be due to incompatibilities between Transformers >=5.0.0 and some models. You can try downgrading to transformers==4.57.1 as a workaround.
sglang-1 | [2026-03-20 03:14:24] Init torch distributed begin.
sglang-1 | [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
sglang-1 | [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
sglang-1 | [Gloo] Rank 0 is connected to 0 peer ranks. Expected number of connected peer ranks is : 0
sglang-1 | [2026-03-20 03:14:24] Init torch distributed ends. elapsed=0.23 s, mem usage=0.03 GB
sglang-1 | [2026-03-20 03:14:25] Load weight begin. avail mem=112.87 GB
sglang-1 | [2026-03-20 03:14:25] Using ModelOptModelLoader due to ModelOpt quantization config.
sglang-1 | [2026-03-20 03:14:25] ModelOptModelLoader: Loading base model...
sglang-1 | [2026-03-20 03:14:25] Model is already quantized, loading directly...
sglang-1 | [2026-03-20 03:14:25] Detected nvfp4 checkpoint. Please note that the format is experimental and subject to change.
sglang-1 | [2026-03-20 03:14:25] Multimodal attention backend not set. Use triton_attn.
sglang-1 | [2026-03-20 03:14:25] Using triton_attn as multimodal attention backend.
sglang-1 | `torch_dtype` is deprecated! Use `dtype` instead!
sglang-1 | [2026-03-20 03:14:25] using attn output gate!
sglang-1 | [2026-03-20 03:14:26] Found local HF snapshot for AxionML/Qwen3.5-35B-A3B-NVFP4 at /models/huggingface/hub/models--AxionML--Qwen3.5-35B-A3B-NVFP4/snapshots/75f5a9d934e5528d372916d8df7376ec52f6801a; skipping download.
sglang-1 | [2026-03-20 03:14:26] No model.safetensors.index.json found in remote.
Loading safetensors checkpoint shards: 0% Completed | 0/1 [00:00<?, ?it/s][2026-03-20 03:14:42] Parameter model.layers.0.linear_attn.in_proj_a.input_scale not found in params_dict
sglang-1 | [2026-03-20 03:14:42] Scheduler hit an exception: Traceback (most recent call last):
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/managers/scheduler.py", line 3424, in run_scheduler_process
sglang-1 | scheduler = Scheduler(
sglang-1 | ^^^^^^^^^^
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/managers/scheduler.py", line 377, in __init__
sglang-1 | self.init_model_worker()
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/managers/scheduler.py", line 612, in init_model_worker
sglang-1 | self.init_tp_model_worker()
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/managers/scheduler.py", line 570, in init_tp_model_worker
sglang-1 | self.tp_worker = TpModelWorker(
sglang-1 | ^^^^^^^^^^^^^^
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/managers/tp_worker.py", line 261, in __init__
sglang-1 | self._init_model_runner()
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/managers/tp_worker.py", line 344, in _init_model_runner
sglang-1 | self._model_runner = ModelRunner(
sglang-1 | ^^^^^^^^^^^^
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/model_executor/model_runner.py", line 422, in __init__
sglang-1 | self.initialize(pre_model_load_memory)
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/model_executor/model_runner.py", line 502, in initialize
sglang-1 | self.load_model()
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/model_executor/model_runner.py", line 1079, in load_model
sglang-1 | self.model = self.loader.load_model(
sglang-1 | ^^^^^^^^^^^^^^^^^^^^^^^
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/model_loader/loader.py", line 2785, in load_model
sglang-1 | return super().load_model(
sglang-1 | ^^^^^^^^^^^^^^^^^^^
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/model_loader/loader.py", line 689, in load_model
sglang-1 | self.load_weights_and_postprocess(
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/model_loader/loader.py", line 698, in load_weights_and_postprocess
sglang-1 | model.load_weights(weights)
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/models/qwen3_5.py", line 1412, in load_weights
sglang-1 | weight_loader(param, loaded_weight)
sglang-1 | File "/sgl-workspace/sglang/python/sglang/srt/layers/linear.py", line 417, in weight_loader
sglang-1 | assert param_data.shape == loaded_weight.shape
sglang-1 | ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
sglang-1 | AssertionError
the same error