| Update README.md | shyamwertyajm | open | 0 | disc |
| Listed on OpenModelMap | duola15 | open | 0 | disc |
| R1 on mobile — we tested the limits | 3morixd | open | 0 | disc |
| New architecture: TemporalMesh Transformer — dynamic kNN graph attention + per-token exit routing, 29.4 PPL at 48% compute | vigneshwar234 | open | 0 | disc |
| test | Luzika01883 | draft | 0 | disc |
| Add EvalEval community eval results | EvalEvalBot | closed | 0 | disc |
| Add LEXam evaluation results | joelniklaus | open | 0 | disc |
| Fix `chat_template` crash when assistant message omits the `content` key | qgallouedec | open | 0 | disc |
| [Feature request] Eliminate pre-attention RMSNorm in MLA-models via scale invariance + weight folding | graefics | open | 0 | disc |
| DeepSeek-R1-1 | usmanashe26 | draft | 0 | disc |
| Using DeepSeek R1 via API Gateway | xujfcn | open | 0 | disc |
| Deepseek R1 Request | bach619 | draft | 0 | disc |
| Add MMLU-Pro evaluation result (84.0) | burtenshaw | open | 0 | disc |
| Add GPQA evaluation result | burtenshaw | open | 0 | disc |
| Add GPQA evaluation results | nielsr | open | 0 | disc |
| Upload internal.py | Ananthusajeev190 | open | 0 | disc |
| Create .eval_results/mmlu_pro.yaml | nielsr | closed | 0 | disc |
| [Alignment Analysis] R1 hallucinates medical false equivalencies unless strictly constrained (Diabetes vs Psychiatry) | felps333 | open | 0 | disc |
| [Hallucination Report] Model hallucinates biological validity in psychiatric analogies | felps333 | open | 0 | disc |
| Set Max Model Length to correct value | chandra-reddy | draft | 0 | disc |
| Update README.md | vivienfanghua | open | 0 | disc |
| Update README.md | Brokersponsor | open | 0 | disc |
| Model's Resource Utilization | kalashshah19 | open | 0 | disc |
| Update README.md | ahmedgg12 | open | 0 | disc |
| Update README.md | Kiokana | closed | 0 | disc |
| Tokenizer difference between deepseek and qwen3 | yangsketch | open | 0 | disc |
| updated version? | CHNtentes | open | 0 | disc |
| 日期出现乱码 | CloudChina | open | 0 | disc |
| test | Bellandry | closed | 0 | disc |
| Update README.md | Xternn | open | 0 | disc |
| dsasd | Axdar | draft | 0 | disc |
| DeepSeek R1 model Bad Cases Genuine User Reviews and Comments Collection | DeepNLP | open | 0 | disc |
| Update README.md | Rudra160113 | open | 0 | disc |
| Update README.md | pezhmansamaniii | open | 0 | disc |
| Update README.md | pezhmansamaniii | open | 0 | disc |
| 🚩 Report: Copyright infringement | king455 | open | 0 | disc |
| Remove "NOTE: Hugging Face's Transformers has not been directly supported yet." in README | andrewrreed | open | 0 | disc |
| Update README.md | rshaikh22 | open | 0 | disc |
| World's Largest Dataset | deleted | open | 0 | disc |
| Request: DOI | yahyaali77824 | open | 0 | disc |
| Knowledge / Training Cutoff of DeepSeek R1 | MengboZhou | open | 0 | disc |
| Update README.md | Rainbowbeast | open | 0 | disc |
| Make config params float to avoid warning in Transformers | Rocketknight1 | open | 0 | disc |
| Update README.md | Brokersponsor | open | 0 | disc |
| Upload 1743103973740.jpg | Diamond01 | closed | 0 | disc |
| Update README.md | mehdi131 | open | 0 | disc |
| Update README.md | YUIHG | open | 0 | disc |
| DeepSeek中o1-1217的数据是哪里来的。我好像没在OpenAI的官方途径找到，谢谢🙏 | 747860199qq | open | 0 | disc |
| Any R1 reasoning researchers looking for samples? | natcolley | open | 0 | disc |
| Update README.md | umar759 | open | 0 | disc |
| Request: DOI | Yenugu12 | open | 0 | disc |
| Create 9889555 | keyi8 | open | 0 | disc |
| Upload 657f0f06e7ea1b09462a7a16_Feedback and evaluation-p-500.png | likhonsheikh | open | 0 | disc |
| lucasPR | lucas2016 | closed | 0 | disc |
| Best practice for R1 models evaluation: Reasoning efficiency and Performance by MATH-Level | wangxingjun778 | open | 0 | disc |
| Производство изобутилена | Marumarymari | closed | 0 | disc |
| DeepSeek R1 full-power version occasionally ends without returning </think>. | yizhiezi | open | 0 | disc |
| deepseek满血版偶现结束没有返回 </think> | yizhiezi | open | 0 | disc |
| Standing at a flag in Netherlands | Sweetstacg | open | 0 | disc |
| Delete Config.json | jana0010 | open | 0 | disc |
| Update README.md | caraanchoa | open | 0 | disc |
| 为助手回答添加 <think>\n> 标签，确保一致性 | REN0430 | open | 0 | disc |
| 为首token加上<think> | REN0430 | closed | 0 | disc |
| fix for transformers 4.49 compatibility | katuni4ka | open | 0 | disc |
| MLLM交流群 | YLHX | open | 0 | disc |
| nothing | whxin | closed | 0 | disc |
| Question about experts select | BXset | open | 0 | disc |
| Hardware Requirements to run the original model - 671B params | EdilCamil | open | 0 | disc |
| Holding paper in hand | Loveyl | open | 0 | disc |
| Standing on the beach | Loveyl | closed | 0 | disc |
| Update config.json | Empolean2640 | open | 0 | disc |
| Regression in Reasoning Tag Output - Missing <think> in Model Responses | divinerapier | open | 0 | disc |
| Delete model.safetensors.index.json | Huggingfaceliaj | open | 0 | disc |
| Unknown quantization type, got fp8 | DenisFavaCerchiaro | open | 0 | disc |
| 如何取消/省略<think></think>过程。 | yech520 | open | 0 | disc |
| Request: DOI | Tamwyn | open | 0 | disc |
| Request: DOI | saathwik | open | 0 | disc |
| Request: DOI | Paulabad | open | 0 | disc |
| Draft model as accelerator for DeepSeek-R1? | inputout | open | 0 | disc |
| Choosing the Appropriate add_noise() Method in diffusers.schedulers and Compatibility with Different Schedulers for Data Generation | llmy | closed | 0 | disc |
| Does R1 support long context (> 4K)? | ghostplant | closed | 0 | disc |
| Deploying production ready service with Unsloth GGUF quants on your AWS account. (4 x L40S) | samagra-tensorfuse | open | 0 | disc |
| 是否可以关注Perplexity推出的“r1-1776”模型？ | yanyihan | open | 0 | disc |
| Just crossed 10,000 likes! | clem | open | 0 | disc |
| mac上面无法下载flash_attn | earlyIsLate | open | 0 | disc |
| Can this model be used for commercial use? | henrycwf | open | 0 | disc |
| 90+ tokens per second for MI300x8 using batch_size = 1 | ghostplant | open | 0 | disc |
| RytryR1 | Rocka01 | open | 0 | disc |
| "aha moment" comment deleted by Perplexity (recovered) | FalconNet | open | 0 | disc |
| 输出乱码 | cell22 | open | 0 | disc |
| 'num_hidden_layers': 61, but layer 62 has weights. | xinhe | open | 0 | disc |
| Upload GTG Breaking every Limit | GTGenesis | open | 0 | disc |
| Update config.json | jana0010 | closed | 0 | disc |
| Upload 1.txt | DuHt | closed | 0 | disc |
| support prefix complete | HuggineAllen | open | 0 | disc |
| Create app.py | SpaceAgeRobotics | open | 0 | disc |
| Create 1 | madevii | open | 0 | disc |
| Brokersponsor | Brokersponsor | open | 0 | disc |
| Update README.md | egegvner | open | 0 | disc |
| deep | zglguilin | closed | 0 | disc |
| Upload IMG_4530.png | Noemie202586 | open | 0 | disc |
| Upload IMG_1745.JPG | Ladib | open | 0 | disc |
| Create Clara | Clblinks | open | 0 | disc |
| If I understand correctly, evaluating MATH-500 requires 64*500 model calls? | Rorschaaaach | open | 0 | disc |
| Request: DOI | Tarush-Appreciate | open | 0 | disc |
| Update README.md | tekno-power | open | 0 | disc |
| Update README.md | Ekimnedops6969 | open | 0 | disc |
| Update README.md | Ekimnedops6969 | closed | 0 | disc |
| Adding <think>\n after chat template will cause vllm to not return reasoning_content (null) when reasoning | kebeliu | closed | 0 | disc |
| Update README.md | MuhammadEhsan | open | 0 | disc |
| Request for Information on Purchasing Reasoning API Key | brahamaandai | open | 0 | disc |
| ssss | DZGT | open | 0 | disc |
| Update model_max_length in tokenizer_config.json | kkokkie2360 | open | 0 | disc |
| Host of the model | henrycwf | open | 0 | disc |
| Lite version for DeepSeek-R1? | haili-tian | open | 0 | disc |
| [Bug] assert not self.training | Gaie | open | 0 | disc |
| fix: update config.json to remove quantazation config | denissellu | closed | 0 | disc |
| Upload IMG_0253.HEIC | rynty | open | 0 | disc |
| Upload comment-sample.xlsx | faham123 | open | 0 | disc |
| non-reasoning data | mccatec | open | 0 | disc |
| 能不能放一些 4bit的权重，现在手里面的卡都不支持FP8 | zhnagchenchne | open | 0 | disc |
| Request: DOI | wxyugbj | closed | 0 | disc |
| For the universe! DeepPhaser.py DeepCoralX.py and DeepSynapse.py | karmikovic | open | 0 | disc |
| Request: Create distill of Mistral Small 24B | Kenshiro-28 | open | 0 | disc |
| which vision model is R1 using for text extraction from image or pdfs. | ashutoshroy02 | open | 0 | disc |
| PRII | bajramani | draft | 0 | disc |
| Request: DOI | Yungchizzy | open | 0 | disc |
| Little brother(s) of big DeepSeek-R1 ? | MrDevolver | open | 0 | disc |
| Upload gugagagaggagagagga.pdf | HahahhahH | open | 0 | disc |
| Tool / Function Calling | smcleod | closed | 0 | disc |
| Change quant_method to bitsandbytes_4bit | ngoc24794 | open | 0 | disc |
| Unknown quantization type | Reewaz321 | open | 0 | disc |
| UPdate config.json | keerthanaOfficial2001 | open | 0 | disc |
| 所以部署一个671B的模型 显存需要多少 有什么基准的硬件配置？ | cena163 | open | 0 | disc |
| nodejs | k1de | open | 0 | disc |
| Upload IMG_20240813_033412.jpg | Amrit12345678 | closed | 0 | disc |
| Distill Compatibility for PC w/ Ryzen 7 Pro 8840HS w/ 780M Graphics 2x32GB RAM 1TB DDR5 SSD | arzx | open | 0 | disc |
| Upload gitattributes.txt | SafeerChalil | open | 0 | disc |
| Introducing Deepseek's TinyZero | DeepSeekModerator | open | 0 | disc |
| Create Kuch v | gamerdowntown | open | 0 | disc |
| Request: DOI | Hassanabbas2975 | open | 0 | disc |
| quantization fp8 error occuring while using pipeline approach or transformer based approach | neethuvm | open | 0 | disc |
| Deepseek-R1 | KudanTao | draft | 0 | disc |
| deepseek-r1 源码中采用 MLA 架构的 KV Cache 压缩存储策略的实现似乎与文中说的不一致，这是为什么？代码中似乎没实现这个大优化 | Darkdust | open | 0 | disc |
| In bedroom | makecash | closed | 0 | disc |
| Eating food in a car | Ayinbaby1313 | open | 0 | disc |
| Him in the hospital | Mrsauce | closed | 0 | disc |
| Update README.md | jungvaclav | open | 0 | disc |
| error while downloading model | heikhama1982 | open | 0 | disc |
| Upload IMG_20250112_172711.jpg | aamir1 | open | 0 | disc |
| help from italy | MMPPIIAA | open | 0 | disc |
| R1 distill to Mistral Small? | nfunctor | open | 0 | disc |
| Running this model on Google Colab? | Zakia | open | 0 | disc |
| 请问下deepseek的同学，能不能train出一个 stable 的 moe model? | tflchina | open | 0 | disc |
| How to download DeepSeek-R1 7B parameters | barqawiz | open | 0 | disc |
| HuggingFace version does NOT use efficient MLA caching | Avelina | open | 0 | disc |
| Found a bug | amalgunatilake | open | 0 | disc |
| Upload IMG-20250115-WA0015.jpg | Ndukadaboss | closed | 0 | disc |
| Let's Give Credit Where It’s Due: Adding Source Links to AI Responses | Munis01 | open | 0 | disc |
| When will this be available in Transformers library? | solwol | open | 0 | disc |
| cannot regenerate (blank respone) | pluhong | open | 0 | disc |
| A Bug using hugging face API | Kevin355 | open | 0 | disc |
| Do we need an authorization access to use this ? | Natwar | open | 0 | disc |
| where is the source code for this Model ? - what does they prodoudly say by open-source models? | tstarksys | open | 0 | disc |
| Hjjjj | Imole2023 | closed | 0 | disc |
| 智王发布deepseek-r1懒人包，解压即用Deepseek-r1 Lazy Package, easy to decompress and use | zwpython | open | 0 | disc |
| model-00078-of-000163.safetensors not marked safe? | aborst | open | 0 | disc |
| Create Dare | Dara996 | open | 0 | disc |
| problem with using serverless inference | manju2345 | open | 0 | disc |
| Some weird sensorship on unsensitive topic. 对非敏感话题的奇怪审查。 | junnanwu | open | 0 | disc |
| Upload dkfoEtm3H4bMcaI0KEJbq.1023.jpeg | luckysalami089 | open | 0 | disc |
| Update README.md | NuoNb | open | 0 | disc |
| 🚩 Report: Ethical issue(s) | Typeofprototype | open | 0 | disc |
| Deepseek-R1 falls: ZW demon redesigns' Nine Birds' Deepseek-R1沦陷：zw魔改版“九只鸟” | zwpython | open | 0 | disc |
| The Magic Token Framework: A Context-Aware, Metacognitive Architecture for Robust AI Decision-Making | Ivan1972 | closed | 0 | disc |
| Consistency, can Deepseek pass?一致性，deepseek能及格吗？ | zwpython | open | 0 | disc |
| Does this model support text insertion (fill in middle)? | AayushShah | open | 0 | disc |
| Thoughts on deepseek-r1. Correct me if I'm wrong | pkms | open | 0 | disc |
| Remove unavailable import | Rocketknight1 | closed | 0 | disc |
| ImportError: cannot import name 'is_torch_greater_or_equal_than_1_13' from 'transformers.pytorch_utils' | bashir-abubakar | open | 0 | disc |
| DeepSeek-R1 // $R1 | DeepSeekAl | closed | 0 | disc |
| e-currency | Zhendaxie | open | 0 | disc |
| Meet PEEPSEEK, the first meme made by DeepSeek r1 | deepseeker3b56 | open | 0 | disc |
| 鲸 Logo transparent | DorianDarko2525 | open | 0 | disc |
| Meet Finley, the Whale of DeepSeek! | deepseekjanus | open | 0 | disc |
| 最近的炒作和硬币 | Chester1111 | open | 0 | disc |
| Office Cat Pippa :3 | deepseekjanus | closed | 0 | disc |
| Official DeepThink Crypto Currency | qwen-llm | open | 0 | disc |
| Congrats, this is the by far the best open source model! Just a few steps until complete domination (feedback) | Dampfinchen | open | 0 | disc |
| deepseek | denizkaya2022 | draft | 0 | disc |
| Modify abbreviations in benchmark images into full name to avoid confusion | karminski | open | 0 | disc |
| Create fork | CwHedgehog | closed | 0 | disc |
| Hizkuntza | Joxio | open | 0 | disc |
| Upload 250d0a04-673e-492b-b97f-a30ebf6a327c.jpeg | Deey22 | closed | 0 | disc |
| How to deploy DeepSeek-R1 witn LMDeploy ? | vansin | open | 0 | disc |
| Qwen | Moukli | open | 0 | disc |
| 使用不带 thinking 的数据集微调时无法正常生成 | HuanLin | open | 0 | disc |
| Use memory to store inactive experts | xm10086 | open | 0 | disc |
| qwen32B蒸馏模型，长度>8k时，预测一定比例乱码，出现<think><think><think><think><think><think> | daniellibin | open | 0 | disc |
| Update LICENSE | town24 | open | 0 | disc |