xinference 1.6.0__py3-none-any.whl → 1.6.1__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Potentially problematic release.
This version of xinference might be problematic. Click here for more details.
- xinference/_version.py +3 -3
- xinference/client/restful/restful_client.py +1 -1
- xinference/conftest.py +0 -7
- xinference/core/media_interface.py +9 -8
- xinference/core/model.py +13 -6
- xinference/core/scheduler.py +1 -10
- xinference/core/worker.py +0 -10
- xinference/model/audio/model_spec.json +53 -1
- xinference/model/audio/model_spec_modelscope.json +57 -1
- xinference/model/embedding/core.py +19 -11
- xinference/model/image/model_spec.json +10 -1
- xinference/model/image/model_spec_modelscope.json +20 -0
- xinference/model/llm/__init__.py +6 -54
- xinference/model/llm/core.py +19 -5
- xinference/model/llm/llama_cpp/core.py +59 -3
- xinference/model/llm/llama_cpp/memory.py +455 -0
- xinference/model/llm/llm_family.json +185 -397
- xinference/model/llm/llm_family.py +88 -16
- xinference/model/llm/llm_family_modelscope.json +199 -421
- xinference/model/llm/llm_family_openmind_hub.json +0 -34
- xinference/model/llm/sglang/core.py +4 -0
- xinference/model/llm/transformers/__init__.py +27 -6
- xinference/model/llm/transformers/chatglm.py +4 -2
- xinference/model/llm/transformers/core.py +49 -28
- xinference/model/llm/transformers/deepseek_v2.py +6 -49
- xinference/model/llm/transformers/gemma3.py +119 -164
- xinference/{thirdparty/omnilmm/train → model/llm/transformers/multimodal}/__init__.py +1 -1
- xinference/model/llm/transformers/{cogagent.py → multimodal/cogagent.py} +58 -95
- xinference/model/llm/transformers/multimodal/core.py +205 -0
- xinference/model/llm/transformers/{deepseek_vl2.py → multimodal/deepseek_vl2.py} +59 -120
- xinference/model/llm/transformers/multimodal/gemma3.py +117 -0
- xinference/model/llm/transformers/{glm4v.py → multimodal/glm4v.py} +57 -93
- xinference/model/llm/transformers/multimodal/intern_vl.py +412 -0
- xinference/model/llm/transformers/{minicpmv26.py → multimodal/minicpmv26.py} +55 -102
- xinference/model/llm/transformers/{ovis2.py → multimodal/ovis2.py} +114 -175
- xinference/model/llm/transformers/{qwen-omni.py → multimodal/qwen-omni.py} +82 -167
- xinference/model/llm/transformers/multimodal/qwen2_audio.py +131 -0
- xinference/model/llm/transformers/{qwen2_vl.py → multimodal/qwen2_vl.py} +224 -256
- xinference/model/llm/transformers/opt.py +4 -2
- xinference/model/llm/transformers/utils.py +6 -37
- xinference/model/llm/vllm/core.py +4 -0
- xinference/model/rerank/core.py +7 -1
- xinference/model/rerank/utils.py +17 -0
- xinference/web/ui/build/asset-manifest.json +3 -3
- xinference/web/ui/build/index.html +1 -1
- xinference/web/ui/build/static/js/main.ddf9eaee.js +3 -0
- xinference/web/ui/build/static/js/main.ddf9eaee.js.map +1 -0
- xinference/web/ui/node_modules/.cache/babel-loader/12e637ed5fa9ca6491b03892b6949c03afd4960fe36ac25744488e7e1982aa19.json +1 -0
- xinference/web/ui/node_modules/.cache/babel-loader/567e49df411efb24425d289bb484758cb57067ca54f8b5c67fe4505f698deb96.json +1 -0
- xinference/web/ui/node_modules/.cache/babel-loader/77ac2665a784e99501ae95d32ef5937837a0439a47e965d291b38e99cb619f5b.json +1 -0
- xinference/web/ui/node_modules/.cache/babel-loader/d4ed4e82bfe69915999ec83f5feaa4301c75ecc6bdf1c78f2d03e4671ecbefc8.json +1 -0
- xinference/web/ui/src/locales/en.json +3 -1
- xinference/web/ui/src/locales/zh.json +3 -1
- {xinference-1.6.0.dist-info → xinference-1.6.1.dist-info}/METADATA +16 -14
- {xinference-1.6.0.dist-info → xinference-1.6.1.dist-info}/RECORD +60 -76
- {xinference-1.6.0.dist-info → xinference-1.6.1.dist-info}/WHEEL +1 -1
- xinference/model/llm/transformers/cogvlm2.py +0 -442
- xinference/model/llm/transformers/cogvlm2_video.py +0 -333
- xinference/model/llm/transformers/deepseek_vl.py +0 -280
- xinference/model/llm/transformers/glm_edge_v.py +0 -213
- xinference/model/llm/transformers/intern_vl.py +0 -526
- xinference/model/llm/transformers/internlm2.py +0 -94
- xinference/model/llm/transformers/minicpmv25.py +0 -193
- xinference/model/llm/transformers/omnilmm.py +0 -132
- xinference/model/llm/transformers/qwen2_audio.py +0 -179
- xinference/model/llm/transformers/qwen_vl.py +0 -360
- xinference/thirdparty/omnilmm/LICENSE +0 -201
- xinference/thirdparty/omnilmm/__init__.py +0 -0
- xinference/thirdparty/omnilmm/chat.py +0 -218
- xinference/thirdparty/omnilmm/constants.py +0 -4
- xinference/thirdparty/omnilmm/conversation.py +0 -332
- xinference/thirdparty/omnilmm/model/__init__.py +0 -1
- xinference/thirdparty/omnilmm/model/omnilmm.py +0 -595
- xinference/thirdparty/omnilmm/model/resampler.py +0 -166
- xinference/thirdparty/omnilmm/model/utils.py +0 -578
- xinference/thirdparty/omnilmm/train/train_utils.py +0 -150
- xinference/thirdparty/omnilmm/utils.py +0 -134
- xinference/web/ui/build/static/js/main.ae579a97.js +0 -3
- xinference/web/ui/build/static/js/main.ae579a97.js.map +0 -1
- xinference/web/ui/node_modules/.cache/babel-loader/2fdc61dcb6a9d1fbcb44be592d0e87d8c3f21297a7327559ef5345665f8343f7.json +0 -1
- xinference/web/ui/node_modules/.cache/babel-loader/3d596a3e8dd6430d7ce81d164e32c31f8d47cfa5f725c328a298754d78563e14.json +0 -1
- xinference/web/ui/node_modules/.cache/babel-loader/5c08e2cd07809ed3e41486b16652253404cbb63a3ff8d0366ee50f57e2413cea.json +0 -1
- xinference/web/ui/node_modules/.cache/babel-loader/8472e58a31720892d534f3febda31f746b25ec4aa60787eef34217b074e67965.json +0 -1
- /xinference/web/ui/build/static/js/{main.ae579a97.js.LICENSE.txt → main.ddf9eaee.js.LICENSE.txt} +0 -0
- {xinference-1.6.0.dist-info → xinference-1.6.1.dist-info}/entry_points.txt +0 -0
- {xinference-1.6.0.dist-info → xinference-1.6.1.dist-info}/licenses/LICENSE +0 -0
- {xinference-1.6.0.dist-info → xinference-1.6.1.dist-info}/top_level.txt +0 -0
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"ast":null,"code":"export var llmAllDataKey=['model_uid','model_name','model_type','model_engine','model_format','model_size_in_billions','quantization','n_worker','n_gpu','n_gpu_layers','replica','request_limits','worker_ip','gpu_idx','download_hub','model_path','reasoning_content','gguf_quantization','gguf_model_path','cpu_offload','peft_model_config','quantization_config','enable_thinking'];export var additionalParameterTipList={'transformers':['torch_dtype','device'],'llama.cpp':['n_ctx','use_mmap','use_mlock'],'vllm':['block_size','gpu_memory_utilization','max_num_seqs','max_model_len','guided_decoding_backend','scheduling_policy','tensor_parallel_size','pipeline_parallel_size','enable_prefix_caching','enable_chunked_prefill','enforce_eager','cpu_offload_gb','disable_custom_all_reduce','limit_mm_per_prompt','model_quantization','mm_processor_kwargs','min_pixels','max_pixels'],'sglang':['mem_fraction_static','attention_reduce_in_fp32','tp_size','dp_size','chunked_prefill_size','cpu_offload_gb','enable_dp_attention','enable_ep_moe'],'mlx':['cache_limit_gb','max_kv_size']};export var quantizationParametersTipList=['load_in_8bit','load_in_4bit','llm_int8_threshold','llm_int8_skip_modules','llm_int8_enable_fp32_cpu_offload','llm_int8_has_fp16_weight','bnb_4bit_compute_dtype','bnb_4bit_quant_type','bnb_4bit_use_double_quant','bnb_4bit_quant_storage'];export var featureModels=[{type:'llm',feature_models:['qwen3','deepseek-v3','deepseek-r1','deepseek-r1-distill-qwen','deepseek-r1-distill-llama','qwen2.5-instruct','qwen2.5-vl-instruct','glm4-0414','QwQ-32B','gemma-3-it']},{type:'embedding',feature_models:['bge-large-zh-v1.5','bge-large-en-v1.5','bge-m3','gte-Qwen2','jina-embeddings-v3']},{type:'rerank',feature_models:[]},{type:'image',feature_models:['FLUX.1-dev','FLUX.1-schnell','sd3.5-large','HunyuanDiT-v1.2','sd3.5-medium']},{type:'audio',feature_models:['CosyVoice2-0.5B','FishSpeech-1.5','F5-TTS','ChatTTS','SenseVoiceSmall','whisper-large-v3']},{type:'video',feature_models:[]}];","map":{"version":3,"names":["llmAllDataKey","additionalParameterTipList","quantizationParametersTipList","featureModels","type","feature_models"],"sources":["/home/runner/work/inference/inference/xinference/web/ui/src/scenes/launch_model/data/data.js"],"sourcesContent":["export const llmAllDataKey = [\n 'model_uid',\n 'model_name',\n 'model_type',\n 'model_engine',\n 'model_format',\n 'model_size_in_billions',\n 'quantization',\n 'n_worker',\n 'n_gpu',\n 'n_gpu_layers',\n 'replica',\n 'request_limits',\n 'worker_ip',\n 'gpu_idx',\n 'download_hub',\n 'model_path',\n 'reasoning_content',\n 'gguf_quantization',\n 'gguf_model_path',\n 'cpu_offload',\n 'peft_model_config',\n 'quantization_config',\n 'enable_thinking',\n]\n\nexport const additionalParameterTipList = {\n 'transformers': ['torch_dtype', 'device'],\n 'llama.cpp': ['n_ctx', 'use_mmap', 'use_mlock'],\n 'vllm': [\n 'block_size',\n 'gpu_memory_utilization',\n 'max_num_seqs',\n 'max_model_len',\n 'guided_decoding_backend',\n 'scheduling_policy',\n 'tensor_parallel_size',\n 'pipeline_parallel_size',\n 'enable_prefix_caching',\n 'enable_chunked_prefill',\n 'enforce_eager',\n 'cpu_offload_gb',\n 'disable_custom_all_reduce',\n 'limit_mm_per_prompt',\n 'model_quantization',\n 'mm_processor_kwargs',\n 'min_pixels',\n 'max_pixels',\n ],\n 'sglang': [\n 'mem_fraction_static',\n 'attention_reduce_in_fp32',\n 'tp_size',\n 'dp_size',\n 'chunked_prefill_size',\n 'cpu_offload_gb',\n 'enable_dp_attention',\n 'enable_ep_moe',\n ],\n 'mlx': ['cache_limit_gb', 'max_kv_size'],\n}\n\nexport const quantizationParametersTipList = [\n 'load_in_8bit',\n 'load_in_4bit',\n 'llm_int8_threshold',\n 'llm_int8_skip_modules',\n 'llm_int8_enable_fp32_cpu_offload',\n 'llm_int8_has_fp16_weight',\n 'bnb_4bit_compute_dtype',\n 'bnb_4bit_quant_type',\n 'bnb_4bit_use_double_quant',\n 'bnb_4bit_quant_storage',\n]\n\nexport const featureModels = [\n {\n type: 'llm',\n feature_models: [\n 'qwen3',\n 'deepseek-v3',\n 'deepseek-r1',\n 'deepseek-r1-distill-qwen',\n 'deepseek-r1-distill-llama',\n 'qwen2.5-instruct',\n 'qwen2.5-vl-instruct',\n 'glm4-0414',\n 'QwQ-32B',\n 'gemma-3-it',\n ],\n },\n {\n type: 'embedding',\n feature_models: [\n 'bge-large-zh-v1.5',\n 'bge-large-en-v1.5',\n 'bge-m3',\n 'gte-Qwen2',\n 'jina-embeddings-v3',\n ],\n },\n {\n type: 'rerank',\n feature_models: [],\n },\n {\n type: 'image',\n feature_models: [\n 'FLUX.1-dev',\n 'FLUX.1-schnell',\n 'sd3.5-large',\n 'HunyuanDiT-v1.2',\n 'sd3.5-medium',\n ],\n },\n {\n type: 'audio',\n feature_models: [\n 'CosyVoice2-0.5B',\n 'FishSpeech-1.5',\n 'F5-TTS',\n 'ChatTTS',\n 'SenseVoiceSmall',\n 'whisper-large-v3',\n ],\n },\n {\n type: 'video',\n feature_models: [],\n },\n]\n"],"mappings":"AAAA,MAAO,IAAM,CAAAA,aAAa,CAAG,CAC3B,WAAW,CACX,YAAY,CACZ,YAAY,CACZ,cAAc,CACd,cAAc,CACd,wBAAwB,CACxB,cAAc,CACd,UAAU,CACV,OAAO,CACP,cAAc,CACd,SAAS,CACT,gBAAgB,CAChB,WAAW,CACX,SAAS,CACT,cAAc,CACd,YAAY,CACZ,mBAAmB,CACnB,mBAAmB,CACnB,iBAAiB,CACjB,aAAa,CACb,mBAAmB,CACnB,qBAAqB,CACrB,iBAAiB,CAClB,CAED,MAAO,IAAM,CAAAC,0BAA0B,CAAG,CACxC,cAAc,CAAE,CAAC,aAAa,CAAE,QAAQ,CAAC,CACzC,WAAW,CAAE,CAAC,OAAO,CAAE,UAAU,CAAE,WAAW,CAAC,CAC/C,MAAM,CAAE,CACN,YAAY,CACZ,wBAAwB,CACxB,cAAc,CACd,eAAe,CACf,yBAAyB,CACzB,mBAAmB,CACnB,sBAAsB,CACtB,wBAAwB,CACxB,uBAAuB,CACvB,wBAAwB,CACxB,eAAe,CACf,gBAAgB,CAChB,2BAA2B,CAC3B,qBAAqB,CACrB,oBAAoB,CACpB,qBAAqB,CACrB,YAAY,CACZ,YAAY,CACb,CACD,QAAQ,CAAE,CACR,qBAAqB,CACrB,0BAA0B,CAC1B,SAAS,CACT,SAAS,CACT,sBAAsB,CACtB,gBAAgB,CAChB,qBAAqB,CACrB,eAAe,CAChB,CACD,KAAK,CAAE,CAAC,gBAAgB,CAAE,aAAa,CACzC,CAAC,CAED,MAAO,IAAM,CAAAC,6BAA6B,CAAG,CAC3C,cAAc,CACd,cAAc,CACd,oBAAoB,CACpB,uBAAuB,CACvB,kCAAkC,CAClC,0BAA0B,CAC1B,wBAAwB,CACxB,qBAAqB,CACrB,2BAA2B,CAC3B,wBAAwB,CACzB,CAED,MAAO,IAAM,CAAAC,aAAa,CAAG,CAC3B,CACEC,IAAI,CAAE,KAAK,CACXC,cAAc,CAAE,CACd,OAAO,CACP,aAAa,CACb,aAAa,CACb,0BAA0B,CAC1B,2BAA2B,CAC3B,kBAAkB,CAClB,qBAAqB,CACrB,WAAW,CACX,SAAS,CACT,YAAY,CAEhB,CAAC,CACD,CACED,IAAI,CAAE,WAAW,CACjBC,cAAc,CAAE,CACd,mBAAmB,CACnB,mBAAmB,CACnB,QAAQ,CACR,WAAW,CACX,oBAAoB,CAExB,CAAC,CACD,CACED,IAAI,CAAE,QAAQ,CACdC,cAAc,CAAE,EAClB,CAAC,CACD,CACED,IAAI,CAAE,OAAO,CACbC,cAAc,CAAE,CACd,YAAY,CACZ,gBAAgB,CAChB,aAAa,CACb,iBAAiB,CACjB,cAAc,CAElB,CAAC,CACD,CACED,IAAI,CAAE,OAAO,CACbC,cAAc,CAAE,CACd,iBAAiB,CACjB,gBAAgB,CAChB,QAAQ,CACR,SAAS,CACT,iBAAiB,CACjB,kBAAkB,CAEtB,CAAC,CACD,CACED,IAAI,CAAE,OAAO,CACbC,cAAc,CAAE,EAClB,CAAC,CACF","ignoreList":[]},"metadata":{},"sourceType":"module","externalDependencies":[]}
|