{"work":{"id":"30da36a4-b9ad-4618-8955-c09232b61343","openalex_id":"https://openalex.org/W4392678661","doi":"10.48550/arxiv.2403.05525","arxiv_id":"2403.05525","raw_key":null,"title":"DeepSeek-VL: Towards Real-World Vision-Language Understanding","authors":null,"authors_text":"Haoyu Lu, Wen Liu, Bo Zhang, Bingxuan Wang, Kai Dong, Bo Liu","year":2024,"venue":"cs.AI","abstract":"We present DeepSeek-VL, an open-source Vision-Language (VL) Model designed for real-world vision and language understanding applications. Our approach is structured around three key dimensions:\n  We strive to ensure our data is diverse, scalable, and extensively covers real-world scenarios including web screenshots, PDFs, OCR, charts, and knowledge-based content, aiming for a comprehensive representation of practical contexts. Further, we create a use case taxonomy from real user scenarios and construct an instruction tuning dataset accordingly. The fine-tuning with this dataset substantially improves the model's user experience in practical applications. Considering efficiency and the demands of most real-world scenarios, DeepSeek-VL incorporates a hybrid vision encoder that efficiently processes high-resolution images (1024 x 1024), while maintaining a relatively low computational overhead. This design choice ensures the model's ability to capture critical semantic and detailed information across various visual tasks. We posit that a proficient Vision-Language Model should, foremost, possess strong language abilities. To ensure the preservation of LLM capabilities during pretraining, we investigate an effective VL pretraining strategy by integrating LLM training from the beginning and carefully managing the competitive dynamics observed between vision and language modalities.\n  The DeepSeek-VL family (both 1.3B and 7B models) showcases superior user experiences as a vision-language chatbot in real-world applications, achieving state-of-the-art or competitive performance across a wide range of visual-language benchmarks at the same model size while maintaining robust performance on language-centric benchmarks. We have made both 1.3B and 7B models publicly accessible to foster innovations based on this foundation model.","external_url":"https://arxiv.org/abs/2403.05525","cited_by_count":46,"metadata_source":"pith","metadata_fetched_at":"2026-08-05T02:28:24.338817+00:00","pith_arxiv_id":"2403.05525","created_at":"2026-05-09T06:40:42.676370+00:00","updated_at":"2026-08-05T02:28:24.338817+00:00","title_quality_ok":true,"display_title":"DeepSeek-VL: Towards Real-World Vision-Language Understanding","render_title":"DeepSeek-VL: Towards Real-World Vision-Language Understanding"},"hub":{"state":{"work_id":"30da36a4-b9ad-4618-8955-c09232b61343","tier":"super_hub","tier_reason":"100+ Pith inbound or 10,000+ external citations","pith_inbound_count":103,"external_cited_by_count":46,"distinct_field_count":8,"first_pith_cited_at":"2024-03-29T17:59:34+00:00","last_pith_cited_at":"2026-07-07T12:41:19+00:00","author_build_status":"needed","summary_status":"needed","contexts_status":"needed","graph_status":"needed","ask_index_status":"needed","reader_status":"not_needed","recognition_status":"not_needed","updated_at":"2026-08-23T05:59:19.258556+00:00","tier_text":"super_hub"},"tier":"super_hub","role_counts":[{"context_role":"background","n":21},{"context_role":"baseline","n":5},{"context_role":"dataset","n":1}],"polarity_counts":[{"context_polarity":"background","n":19},{"context_polarity":"baseline","n":5},{"context_polarity":"support","n":1},{"context_polarity":"unclear","n":1},{"context_polarity":"use_dataset","n":1}],"runs":{"ask_index":{"job_type":"ask_index","status":"succeeded","result":{"title":"DeepSeek-VL: Towards Real-World Vision-Language Understanding","claims":[{"claim_text":"We present DeepSeek-VL, an open-source Vision-Language (VL) Model designed for real-world vision and language understanding applications. Our approach is structured around three key dimensions:\n  We strive to ensure our data is diverse, scalable, and extensively covers real-world scenarios including web screenshots, PDFs, OCR, charts, and knowledge-based content, aiming for a comprehensive representation of practical contexts. Further, we create a use case taxonomy from real user scenarios and construct an instruction tuning dataset accordingly. The fine-tuning with this dataset substantially ","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"\"Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks,\" in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 24 185-24 198. [145] H. Lu, W. Liu, B. Zhang, B. Wang, K. Dong, B. Liu, J. Sun, T. Ren, Z. Li, H. Yang et al., \"Deepseek-vl: towards real-world vision-language understanding,\" arXiv preprint arXiv:2403.05525, 2024. [146] Z. Liu, L. Zhu, B. Shi, Z. Zhang, Y . Lou, S. Yang, H. Xi, S. Cao, Y . Gu, D. Liet al","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"0 K) LVIS-Instruct4V [132] (220.0 K) Language (23.8%) GeomVerse [65] (9.0 K) ChartQA [97] (28.0 K) LLaVA150K [83] (150.0 K) OpenOrca [79] (994.0 K) MathVision [133] (3.0 K) IconQA [90] (27.0 K) VisualGenome [70] (86.0 K) MathInstruct [143] (262.0 K) Inter-GPS [91] (1.0 K) Chart2Text [63] (26.0 K) VQAv2 [51] (83.0 K) OrcaMath [100] (200.0 K) TQA [5] (1.0 K) TabMWP [89] (23.0 K) GPT4V Rewritten (77.0 K) WizardCoder [94] (143.0 K) Science (2.9%) TextCaps [120] (22.0 K) GQA [58] (72.0 K) OpenCodeInt","claim_type":"dataset","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"This is typically achieved by connecting a pre-trained vision encoder [9] to the LLM via a lightweight adapter [24, 25]. This architecture, particularly when refined with large-scalevisual instruction tuning[24], has proven highly effective. It has produced a wave of influentialopen- source models-such as LLaV A-1.5 [25], Qwen-VL [2], and Deepseek-VL [27]-as well as cutting-edge proprietary systems like OpenAI's GPT models and Google's Gemini [36]. These models demonstrate unprecedented conversa","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Model Type LLM Tower MMMU MMMU-Pro MMMU-Pro MMStar MME SeedB MMB MathVerse MathVistaval standard vision test cog./perp. image en-dev mini-vision testmini ShareGPT4V[77] AR Vicuna-7B - - - - 376/1567 69.7 68.8 - -Cambrian-1[78] AR LLaMA3-8B42.7 - - - -/1547 74.7 75.9 - 49.0LLaV A[1] AR Vicuna-7B - - - - -/809 37.0 38.7 - -LLaV A-1.5[2] AR Vicuna-7B - - - - -/1510 66.1 64.3 - -Qwen2-VL[5] AR Qwen2-7B 54.1 43.5 - 60.7 - - - - 58.2DeepSeek-VL[79] AR DeepSeek-7B36.6 - - - - 70.4 73.2 - -DeepSeek-VL2[","claim_type":"baseline","confidence":0.85,"evidence_strength":"citation_context"},{"claim_text":"end, including the vision encoder and the image-to-text projection module. As a result, a line of work focuses on designing more expressive projectors [13,49,56], for example by aggregating multi-layer features from the vision encoder before being fed into the LLM [14, 47], while other studies explore the use of multiple vision encoders to enrich visual representations [5, 40, 54, 61, 65, 66]. Another line of work identifies visual bottleneck not in visual representation quality but in the utili","claim_type":"background","confidence":0.85,"evidence_strength":"citation_context"},{"claim_text":"tures have been widely applied, achieving strong performance ac- ross traditional hyperspectral tasks [ 60]. However, most existing studies focus on task-specific models, while rare attention has been given to how HSI data can be inter- preted and utilized by general MLLMs. Raw HSI cubes are typically incompatible with current MLLMs input formats, making repre- sentation adaptation essential [ 39]. Potential approaches include textual descriptions, pseudo-color visualizations [ 25], or reduced- ","claim_type":"background","confidence":0.85,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks DeepSeek-VL: Towards Real-World Vision-Language Understanding because it crossed a citation-hub threshold. Current citing contexts most often use it as background evidence (21 contexts).","role_counts":[{"n":21,"context_role":"background"},{"n":5,"context_role":"baseline"},{"n":1,"context_role":"dataset"}]},"error":null,"updated_at":"2026-07-04T06:46:37.980843+00:00"},"author_expand":{"job_type":"author_expand","status":"succeeded","result":{"authors_linked":[{"id":"97e6ac46-1884-4b88-955f-7ec611cab8f2","orcid":null,"display_name":"Haoyu Lu"},{"id":"4459ebab-a125-45b5-b329-26e3261e6968","orcid":null,"display_name":"Wen Liu"},{"id":"89db5aae-04a6-44a6-83d9-c676c900cf8c","orcid":null,"display_name":"Bo Zhang"},{"id":"1319a98c-3288-417b-86b2-19490b8cdcb2","orcid":null,"display_name":"Bingxuan Wang"},{"id":"f93ab3fb-8ba4-442b-b38b-63a91d981d58","orcid":null,"display_name":"Kai Dong"},{"id":"d23ba457-7f27-40ff-b81f-5e71cbdfd3cc","orcid":null,"display_name":"Bo Liu"}]},"error":null,"updated_at":"2026-07-04T06:46:37.586259+00:00"},"context_extract":{"job_type":"context_extract","status":"succeeded","result":{"enqueued_papers":25},"error":null,"updated_at":"2026-05-14T18:19:40.033907+00:00"},"graph_features":{"job_type":"graph_features","status":"succeeded","result":{"co_cited":[{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":14},{"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","shared_citers":12},{"title":"Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution","work_id":"8abcfe4f-e0fb-44b7-9123-448fac95f90a","shared_citers":11},{"title":"Gemini: A Family of Highly Capable Multimodal Models","work_id":"83f7c85b-3f11-450f-ac0c-64d9745220b2","shared_citers":10},{"title":"MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models","work_id":"806d2e73-71b3-4d56-87e0-39d571cc15d6","shared_citers":10},{"title":"Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context","work_id":"80e3e977-f1bb-4c83-8d0c-1ab0a0c5c3f1","shared_citers":8},{"title":"LLaVA-OneVision: Easy Visual Task Transfer","work_id":"f5f2452b-f2a9-49ac-b38d-c76e18cdfe49","shared_citers":8},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":8},{"title":"Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond","work_id":"cbc2bb21-b6bb-46c0-80bf-107e195ffe10","shared_citers":8},{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":8},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":7},{"title":"MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models","work_id":"a7e3a737-e007-42bc-be89-c4d34c5ee071","shared_citers":7},{"title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","work_id":"008df105-2fdd-45d8-857a-8e35868aecb6","shared_citers":6},{"title":"InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency","work_id":"b8f5e260-fff5-444e-bcf5-2c42cfefd83d","shared_citers":6},{"title":"Kosmos-2: Grounding Multimodal Large Language Models to the World","work_id":"46e7f9e9-24c6-49af-b7d5-96159fa6f443","shared_citers":6},{"title":"Mitigating Hallucination in Large Multi-Modal Models via Robust Instruction Tuning","work_id":"ba8e8164-e47f-42d6-83ad-696cb57ee79a","shared_citers":6},{"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","shared_citers":5},{"title":"Gemma 3 Technical Report","work_id":"f93e08bf-9e96-409b-8ac6-b8385fd17fd7","shared_citers":5},{"title":"InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models","work_id":"fe8637aa-12bc-4434-8d36-9f57b5eebcbe","shared_citers":5},{"title":"Llama 2: Open Foundation and Fine-Tuned Chat Models","work_id":"68a5177f-d644-44c1-bd4f-4e5278c22f5d","shared_citers":5},{"title":"MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts","work_id":"e22c3789-9e71-4242-b6ea-3e60e06e2b66","shared_citers":5},{"title":"MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities","work_id":"7f3bac41-a0a5-4a7a-bfd2-526b616db745","shared_citers":5},{"title":"Qwen Technical Report","work_id":"bb1fd52f-6b2f-437c-9516-37bdf6eb9be8","shared_citers":5},{"title":"Shikra: Unleashing Multimodal LLM's Referential Dialogue Magic","work_id":"44525076-312a-4259-b79c-134cd7eeb297","shared_citers":5}],"time_series":[{"n":5,"year":2024},{"n":28,"year":2026}],"dependency_candidates":[]},"error":null,"updated_at":"2026-05-14T18:19:57.009005+00:00"},"identity_refresh":{"job_type":"identity_refresh","status":"succeeded","result":{"items":[{"title":"Qwen3 Technical Report","outcome":"unchanged","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","resolver":"local_arxiv","confidence":0.98,"old_work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e"}],"counts":{"fixed":0,"merged":0,"unchanged":1,"quarantined":0,"needs_external_resolution":0},"errors":[],"attempted":1},"error":null,"updated_at":"2026-05-14T18:19:44.462614+00:00"},"role_polarity":{"job_type":"role_polarity","status":"succeeded","result":{"title":"DeepSeek-VL: Towards Real-World Vision-Language Understanding","claims":[{"claim_text":"We present DeepSeek-VL, an open-source Vision-Language (VL) Model designed for real-world vision and language understanding applications. Our approach is structured around three key dimensions:\n  We strive to ensure our data is diverse, scalable, and extensively covers real-world scenarios including web screenshots, PDFs, OCR, charts, and knowledge-based content, aiming for a comprehensive representation of practical contexts. Further, we create a use case taxonomy from real user scenarios and construct an instruction tuning dataset accordingly. The fine-tuning with this dataset substantially ","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"\"Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks,\" in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 24 185-24 198. [145] H. Lu, W. Liu, B. Zhang, B. Wang, K. Dong, B. Liu, J. Sun, T. Ren, Z. Li, H. Yang et al., \"Deepseek-vl: towards real-world vision-language understanding,\" arXiv preprint arXiv:2403.05525, 2024. [146] Z. Liu, L. Zhu, B. Shi, Z. Zhang, Y . Lou, S. Yang, H. Xi, S. Cao, Y . Gu, D. Liet al","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"0 K) LVIS-Instruct4V [132] (220.0 K) Language (23.8%) GeomVerse [65] (9.0 K) ChartQA [97] (28.0 K) LLaVA150K [83] (150.0 K) OpenOrca [79] (994.0 K) MathVision [133] (3.0 K) IconQA [90] (27.0 K) VisualGenome [70] (86.0 K) MathInstruct [143] (262.0 K) Inter-GPS [91] (1.0 K) Chart2Text [63] (26.0 K) VQAv2 [51] (83.0 K) OrcaMath [100] (200.0 K) TQA [5] (1.0 K) TabMWP [89] (23.0 K) GPT4V Rewritten (77.0 K) WizardCoder [94] (143.0 K) Science (2.9%) TextCaps [120] (22.0 K) GQA [58] (72.0 K) OpenCodeInt","claim_type":"dataset","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"This is typically achieved by connecting a pre-trained vision encoder [9] to the LLM via a lightweight adapter [24, 25]. This architecture, particularly when refined with large-scalevisual instruction tuning[24], has proven highly effective. It has produced a wave of influentialopen- source models-such as LLaV A-1.5 [25], Qwen-VL [2], and Deepseek-VL [27]-as well as cutting-edge proprietary systems like OpenAI's GPT models and Google's Gemini [36]. These models demonstrate unprecedented conversa","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Model Type LLM Tower MMMU MMMU-Pro MMMU-Pro MMStar MME SeedB MMB MathVerse MathVistaval standard vision test cog./perp. image en-dev mini-vision testmini ShareGPT4V[77] AR Vicuna-7B - - - - 376/1567 69.7 68.8 - -Cambrian-1[78] AR LLaMA3-8B42.7 - - - -/1547 74.7 75.9 - 49.0LLaV A[1] AR Vicuna-7B - - - - -/809 37.0 38.7 - -LLaV A-1.5[2] AR Vicuna-7B - - - - -/1510 66.1 64.3 - -Qwen2-VL[5] AR Qwen2-7B 54.1 43.5 - 60.7 - - - - 58.2DeepSeek-VL[79] AR DeepSeek-7B36.6 - - - - 70.4 73.2 - -DeepSeek-VL2[","claim_type":"baseline","confidence":0.85,"evidence_strength":"citation_context"},{"claim_text":"end, including the vision encoder and the image-to-text projection module. As a result, a line of work focuses on designing more expressive projectors [13,49,56], for example by aggregating multi-layer features from the vision encoder before being fed into the LLM [14, 47], while other studies explore the use of multiple vision encoders to enrich visual representations [5, 40, 54, 61, 65, 66]. Another line of work identifies visual bottleneck not in visual representation quality but in the utili","claim_type":"background","confidence":0.85,"evidence_strength":"citation_context"},{"claim_text":"tures have been widely applied, achieving strong performance ac- ross traditional hyperspectral tasks [ 60]. However, most existing studies focus on task-specific models, while rare attention has been given to how HSI data can be inter- preted and utilized by general MLLMs. Raw HSI cubes are typically incompatible with current MLLMs input formats, making repre- sentation adaptation essential [ 39]. Potential approaches include textual descriptions, pseudo-color visualizations [ 25], or reduced- ","claim_type":"background","confidence":0.85,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks DeepSeek-VL: Towards Real-World Vision-Language Understanding because it crossed a citation-hub threshold. Current citing contexts most often use it as background evidence (21 contexts).","role_counts":[{"n":21,"context_role":"background"},{"n":5,"context_role":"baseline"},{"n":1,"context_role":"dataset"}]},"error":null,"updated_at":"2026-07-04T06:46:38.421733+00:00"},"summary_claims":{"job_type":"summary_claims","status":"succeeded","result":{"title":"DeepSeek-VL: Towards Real-World Vision-Language Understanding","claims":[{"claim_text":"We present DeepSeek-VL, an open-source Vision-Language (VL) Model designed for real-world vision and language understanding applications. Our approach is structured around three key dimensions:\n  We strive to ensure our data is diverse, scalable, and extensively covers real-world scenarios including web screenshots, PDFs, OCR, charts, and knowledge-based content, aiming for a comprehensive representation of practical contexts. Further, we create a use case taxonomy from real user scenarios and construct an instruction tuning dataset accordingly. The fine-tuning with this dataset substantially ","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks DeepSeek-VL: Towards Real-World Vision-Language Understanding because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T18:19:25.711147+00:00"}},"summary":{"title":"DeepSeek-VL: Towards Real-World Vision-Language Understanding","claims":[{"claim_text":"We present DeepSeek-VL, an open-source Vision-Language (VL) Model designed for real-world vision and language understanding applications. Our approach is structured around three key dimensions:\n  We strive to ensure our data is diverse, scalable, and extensively covers real-world scenarios including web screenshots, PDFs, OCR, charts, and knowledge-based content, aiming for a comprehensive representation of practical contexts. Further, we create a use case taxonomy from real user scenarios and construct an instruction tuning dataset accordingly. The fine-tuning with this dataset substantially ","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks DeepSeek-VL: Towards Real-World Vision-Language Understanding because it crossed a citation-hub threshold.","role_counts":[]},"graph":{"co_cited":[{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":14},{"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","shared_citers":12},{"title":"Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution","work_id":"8abcfe4f-e0fb-44b7-9123-448fac95f90a","shared_citers":11},{"title":"Gemini: A Family of Highly Capable Multimodal Models","work_id":"83f7c85b-3f11-450f-ac0c-64d9745220b2","shared_citers":10},{"title":"MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models","work_id":"806d2e73-71b3-4d56-87e0-39d571cc15d6","shared_citers":10},{"title":"Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context","work_id":"80e3e977-f1bb-4c83-8d0c-1ab0a0c5c3f1","shared_citers":8},{"title":"LLaVA-OneVision: Easy Visual Task Transfer","work_id":"f5f2452b-f2a9-49ac-b38d-c76e18cdfe49","shared_citers":8},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":8},{"title":"Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond","work_id":"cbc2bb21-b6bb-46c0-80bf-107e195ffe10","shared_citers":8},{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":8},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":7},{"title":"MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models","work_id":"a7e3a737-e007-42bc-be89-c4d34c5ee071","shared_citers":7},{"title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","work_id":"008df105-2fdd-45d8-857a-8e35868aecb6","shared_citers":6},{"title":"InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency","work_id":"b8f5e260-fff5-444e-bcf5-2c42cfefd83d","shared_citers":6},{"title":"Kosmos-2: Grounding Multimodal Large Language Models to the World","work_id":"46e7f9e9-24c6-49af-b7d5-96159fa6f443","shared_citers":6},{"title":"Mitigating Hallucination in Large Multi-Modal Models via Robust Instruction Tuning","work_id":"ba8e8164-e47f-42d6-83ad-696cb57ee79a","shared_citers":6},{"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","shared_citers":5},{"title":"Gemma 3 Technical Report","work_id":"f93e08bf-9e96-409b-8ac6-b8385fd17fd7","shared_citers":5},{"title":"InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models","work_id":"fe8637aa-12bc-4434-8d36-9f57b5eebcbe","shared_citers":5},{"title":"Llama 2: Open Foundation and Fine-Tuned Chat Models","work_id":"68a5177f-d644-44c1-bd4f-4e5278c22f5d","shared_citers":5},{"title":"MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts","work_id":"e22c3789-9e71-4242-b6ea-3e60e06e2b66","shared_citers":5},{"title":"MM-Vet: Evaluating Large Multimodal Models for Integrated Capabilities","work_id":"7f3bac41-a0a5-4a7a-bfd2-526b616db745","shared_citers":5},{"title":"Qwen Technical Report","work_id":"bb1fd52f-6b2f-437c-9516-37bdf6eb9be8","shared_citers":5},{"title":"Shikra: Unleashing Multimodal LLM's Referential Dialogue Magic","work_id":"44525076-312a-4259-b79c-134cd7eeb297","shared_citers":5}],"time_series":[{"n":5,"year":2024},{"n":28,"year":2026}],"dependency_candidates":[]},"authors":[{"id":"1319a98c-3288-417b-86b2-19490b8cdcb2","orcid":null,"display_name":"Bingxuan Wang","source":"manual","import_confidence":0.72},{"id":"d23ba457-7f27-40ff-b81f-5e71cbdfd3cc","orcid":null,"display_name":"Bo Liu","source":"manual","import_confidence":0.72},{"id":"89db5aae-04a6-44a6-83d9-c676c900cf8c","orcid":null,"display_name":"Bo Zhang","source":"manual","import_confidence":0.72},{"id":"97e6ac46-1884-4b88-955f-7ec611cab8f2","orcid":null,"display_name":"Haoyu Lu","source":"manual","import_confidence":0.72},{"id":"f93ab3fb-8ba4-442b-b38b-63a91d981d58","orcid":null,"display_name":"Kai Dong","source":"manual","import_confidence":0.72},{"id":"4459ebab-a125-45b5-b329-26e3261e6968","orcid":null,"display_name":"Wen Liu","source":"manual","import_confidence":0.72}]}}