From d084318b958ab94e95ade4fe5c86281907ac3e4d Mon Sep 17 00:00:00 2001 From: Minamiyama Date: Tue, 8 Apr 2025 00:41:32 +0800 Subject: [PATCH 1/2] add model cards --- xinference/model/llm/llm_family.json | 47 +++++++++++++++++ .../model/llm/llm_family_modelscope.json | 50 +++++++++++++++++++ 2 files changed, 97 insertions(+) diff --git a/xinference/model/llm/llm_family.json b/xinference/model/llm/llm_family.json index ad7db9aa48..cc2e2066f2 100644 --- a/xinference/model/llm/llm_family.json +++ b/xinference/model/llm/llm_family.json @@ -10874,5 +10874,52 @@ "stop": [ "<｜end▁of▁sentence｜>" ] + }, + { + "version": 1, + "context_length": 131072, + "model_name": "mPLUG-Owl3", + "model_lang": [ + "en", + "zh" + ], + "model_ability": [ + "chat", + "vision" + ], + "model_description": "mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models", + "model_specs": [ + { + "model_format": "pytorch", + "model_size_in_billions": 1, + "quantizations": [ + "none" + ], + "model_id": "mPLUG/mPLUG-Owl3-1B-241014" + }, + { + "model_format": "pytorch", + "model_size_in_billions": 2, + "quantizations": [ + "none" + ], + "model_id": "mPLUG/mPLUG-Owl3-2B-241014" + }, + { + "model_format": "pytorch", + "model_size_in_billions": 7, + "quantizations": [ + "none" + ], + "model_id": "mPLUG/mPLUG-Owl3-7B-241101" + } + ], + "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}", + "stop_token_ids": [ + 151645 + ], + "stop": [ + "<|im_end|>" + ] } ] diff --git a/xinference/model/llm/llm_family_modelscope.json b/xinference/model/llm/llm_family_modelscope.json index 01cc1cda9a..871b9b9afa 100644 --- a/xinference/model/llm/llm_family_modelscope.json +++ b/xinference/model/llm/llm_family_modelscope.json @@ -8544,5 +8544,55 @@ "stop": [ "<｜end▁of▁sentence｜>" ] + }, + { + "version": 1, + "context_length": 131072, + "model_name": "mPLUG-Owl3", + "model_lang": [ + "en", + "zh" + ], + "model_ability": [ + "chat", + "vision" + ], + "model_description": "mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models", + "model_specs": [ + { + "model_format": "pytorch", + "model_size_in_billions": 1, + "quantizations": [ + "none" + ], + "model_id": "iic/mPLUG-Owl3-1B-241014", + "model_hub": "modelscope" + }, + { + "model_format": "pytorch", + "model_size_in_billions": 2, + "quantizations": [ + "none" + ], + "model_id": "iic/mPLUG-Owl3-2B-241014", + "model_hub": "modelscope" + }, + { + "model_format": "pytorch", + "model_size_in_billions": 7, + "quantizations": [ + "none" + ], + "model_id": "iic/mPLUG-Owl3-7B-241101", + "model_hub": "modelscope" + } + ], + "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}", + "stop_token_ids": [ + 151645 + ], + "stop": [ + "<|im_end|>" + ] } ] From b405a22a65d50e34af1c32c7e02d7c0c75e0cdbb Mon Sep 17 00:00:00 2001 From: Minamiyama Date: Thu, 24 Apr 2025 08:57:38 +0800 Subject: [PATCH 2/2] reformatted --- xinference/model/llm/llm_family.json | 48 +++++++++---------- .../model/llm/llm_family_modelscope.json | 48 +++++++++---------- 2 files changed, 48 insertions(+), 48 deletions(-) diff --git a/xinference/model/llm/llm_family.json b/xinference/model/llm/llm_family.json index cc2e2066f2..1d1820274c 100644 --- a/xinference/model/llm/llm_family.json +++ b/xinference/model/llm/llm_family.json @@ -10830,8 +10830,8 @@ }, { "version": 1, - "context_length": 4096, - "model_name": "deepseek-vl2", + "context_length": 131072, + "model_name": "mPLUG-Owl3", "model_lang": [ "en", "zh" @@ -10840,45 +10840,45 @@ "chat", "vision" ], - "model_description": "DeepSeek-VL2, an advanced series of large Mixture-of-Experts (MoE) Vision-Language Models that significantly improves upon its predecessor, DeepSeek-VL. DeepSeek-VL2 demonstrates superior capabilities across various tasks, including but not limited to visual question answering, optical character recognition, document/table/chart understanding, and visual grounding.", + "model_description": "mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models", "model_specs": [ { "model_format": "pytorch", - "model_size_in_billions": 27, + "model_size_in_billions": 1, "quantizations": [ "none" ], - "model_id": "deepseek-ai/deepseek-vl2" + "model_id": "mPLUG/mPLUG-Owl3-1B-241014" }, { "model_format": "pytorch", - "model_size_in_billions": 16, + "model_size_in_billions": 2, "quantizations": [ "none" ], - "model_id": "deepseek-ai/deepseek-vl2-small" + "model_id": "mPLUG/mPLUG-Owl3-2B-241014" }, { "model_format": "pytorch", - "model_size_in_billions": 3, + "model_size_in_billions": 7, "quantizations": [ "none" ], - "model_id": "deepseek-ai/deepseek-vl2-tiny" + "model_id": "mPLUG/mPLUG-Owl3-7B-241101" } ], - "chat_template": "", + "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}", "stop_token_ids": [ - 1 + 151645 ], "stop": [ - "<｜end▁of▁sentence｜>" + "<|im_end|>" ] }, { "version": 1, - "context_length": 131072, - "model_name": "mPLUG-Owl3", + "context_length": 4096, + "model_name": "deepseek-vl2", "model_lang": [ "en", "zh" @@ -10887,39 +10887,39 @@ "chat", "vision" ], - "model_description": "mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models", + "model_description": "DeepSeek-VL2, an advanced series of large Mixture-of-Experts (MoE) Vision-Language Models that significantly improves upon its predecessor, DeepSeek-VL. DeepSeek-VL2 demonstrates superior capabilities across various tasks, including but not limited to visual question answering, optical character recognition, document/table/chart understanding, and visual grounding.", "model_specs": [ { "model_format": "pytorch", - "model_size_in_billions": 1, + "model_size_in_billions": 27, "quantizations": [ "none" ], - "model_id": "mPLUG/mPLUG-Owl3-1B-241014" + "model_id": "deepseek-ai/deepseek-vl2" }, { "model_format": "pytorch", - "model_size_in_billions": 2, + "model_size_in_billions": 16, "quantizations": [ "none" ], - "model_id": "mPLUG/mPLUG-Owl3-2B-241014" + "model_id": "deepseek-ai/deepseek-vl2-small" }, { "model_format": "pytorch", - "model_size_in_billions": 7, + "model_size_in_billions": 3, "quantizations": [ "none" ], - "model_id": "mPLUG/mPLUG-Owl3-7B-241101" + "model_id": "deepseek-ai/deepseek-vl2-tiny" } ], - "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}", + "chat_template": "", "stop_token_ids": [ - 151645 + 1 ], "stop": [ - "<|im_end|>" + "<｜end▁of▁sentence｜>" ] } ] diff --git a/xinference/model/llm/llm_family_modelscope.json b/xinference/model/llm/llm_family_modelscope.json index 871b9b9afa..76f82a0bde 100644 --- a/xinference/model/llm/llm_family_modelscope.json +++ b/xinference/model/llm/llm_family_modelscope.json @@ -8497,8 +8497,8 @@ }, { "version": 1, - "context_length": 4096, - "model_name": "deepseek-vl2", + "context_length": 131072, + "model_name": "mPLUG-Owl3", "model_lang": [ "en", "zh" @@ -8507,48 +8507,48 @@ "chat", "vision" ], - "model_description": "DeepSeek-VL2, an advanced series of large Mixture-of-Experts (MoE) Vision-Language Models that significantly improves upon its predecessor, DeepSeek-VL. DeepSeek-VL2 demonstrates superior capabilities across various tasks, including but not limited to visual question answering, optical character recognition, document/table/chart understanding, and visual grounding.", + "model_description": "mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models", "model_specs": [ { "model_format": "pytorch", - "model_size_in_billions": 27, + "model_size_in_billions": 1, "quantizations": [ "none" ], - "model_id": "deepseek-ai/deepseek-vl2", + "model_id": "iic/mPLUG-Owl3-1B-241014", "model_hub": "modelscope" }, { "model_format": "pytorch", - "model_size_in_billions": 16, + "model_size_in_billions": 2, "quantizations": [ "none" ], - "model_id": "deepseek-ai/deepseek-vl2-small", + "model_id": "iic/mPLUG-Owl3-2B-241014", "model_hub": "modelscope" }, { "model_format": "pytorch", - "model_size_in_billions": 3, + "model_size_in_billions": 7, "quantizations": [ "none" ], - "model_id": "deepseek-ai/deepseek-vl2-tiny", + "model_id": "iic/mPLUG-Owl3-7B-241101", "model_hub": "modelscope" } ], - "chat_template": "", + "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}", "stop_token_ids": [ - 1 + 151645 ], "stop": [ - "<｜end▁of▁sentence｜>" + "<|im_end|>" ] }, { "version": 1, - "context_length": 131072, - "model_name": "mPLUG-Owl3", + "context_length": 4096, + "model_name": "deepseek-vl2", "model_lang": [ "en", "zh" @@ -8557,42 +8557,42 @@ "chat", "vision" ], - "model_description": "mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models", + "model_description": "DeepSeek-VL2, an advanced series of large Mixture-of-Experts (MoE) Vision-Language Models that significantly improves upon its predecessor, DeepSeek-VL. DeepSeek-VL2 demonstrates superior capabilities across various tasks, including but not limited to visual question answering, optical character recognition, document/table/chart understanding, and visual grounding.", "model_specs": [ { "model_format": "pytorch", - "model_size_in_billions": 1, + "model_size_in_billions": 27, "quantizations": [ "none" ], - "model_id": "iic/mPLUG-Owl3-1B-241014", + "model_id": "deepseek-ai/deepseek-vl2", "model_hub": "modelscope" }, { "model_format": "pytorch", - "model_size_in_billions": 2, + "model_size_in_billions": 16, "quantizations": [ "none" ], - "model_id": "iic/mPLUG-Owl3-2B-241014", + "model_id": "deepseek-ai/deepseek-vl2-small", "model_hub": "modelscope" }, { "model_format": "pytorch", - "model_size_in_billions": 7, + "model_size_in_billions": 3, "quantizations": [ "none" ], - "model_id": "iic/mPLUG-Owl3-7B-241101", + "model_id": "deepseek-ai/deepseek-vl2-tiny", "model_hub": "modelscope" } ], - "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}", + "chat_template": "", "stop_token_ids": [ - 151645 + 1 ], "stop": [ - "<|im_end|>" + "<｜end▁of▁sentence｜>" ] } ]