https://huggingface.co/papers/2504.02441
Covers what is used specifically in LLMs: external memory, KV-cache methods, parameter-based approaches, and hidden-state models, with concrete techniques for storage, retrieval, and compression.

\n\n

MemOS: A Memory OS for AI System -> https://huggingface.co/papers/2507.03724
Introduces MemOS, a memory operating system for LLMs that unifies parameter, activation, and external memories, for explicit memory management, lower training and inference costs, and more updatable knowledge across interactions

\n

\n

MemEvolve: Meta-Evolution of Agent Memory Systems -> https://huggingface.co/papers/2512.18746
MemEvolve is a framework that shows how to jointly adapt agent experience and memory architecture. It also presents EvolveLab, a modular codebase for comparing memory designs, showing improved performance and transfer across tasks and LLMs

\n

\n\n"}},{"time":"2025-12-28T12:52:48.480Z","user":"Kseniase","userAvatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","type":"social-post","socialPost":{"slug":"637856294883643","content":[{"type":"text","value":"What we learned about memory in 2025: 8 comprehensive resources","raw":"What we learned about memory in 2025: 8 comprehensive resources"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"If models forget everything, how can they be reliable? AI systems need to remember past interactions, update knowledge, stay consistent over time, and work beyond a single prompt. That's why many start to talk more about memory in AI. ","raw":"If models forget everything, how can they be reliable? AI systems need to remember past interactions, update knowledge, stay consistent over time, and work beyond a single prompt. That's why many start to talk more about memory in AI. "},{"type":"new_line","raw":"\n"},{"type":"text","value":"Here’s a useful set of studies and videos on where AI memory stands today:","raw":"Here’s a useful set of studies and videos on where AI memory stands today:"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"1. ","raw":"1. "},{"type":"resource","resource":{"type":"paper","id":"2512.13564"},"url":"https://huggingface.co/papers/2512.13564","raw":"https://huggingface.co/papers/2512.13564","label":"Memory in the Age of AI Agents (2512.13564)"},{"type":"new_line","raw":"\n"},{"type":"text","value":"A great survey that organizes agent memory research. It gives concrete taxonomies across memory form, function, and dynamics, summarizes benchmarks, frameworks, and emerging directions for building systematic agent memory systems","raw":"A great survey that organizes agent memory research. It gives concrete taxonomies across memory form, function, and dynamics, summarizes benchmarks, frameworks, and emerging directions for building systematic agent memory systems"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"2.When Will We Give AI True Memory? A conversation with Edo Liberty, CEO and founder @ Pinecone -> ","raw":"2.When Will We Give AI True Memory? A conversation with Edo Liberty, CEO and founder @ Pinecone -> "},{"type":"link","href":"https://youtu.be/ITbwVFZYepc?si=_lAbRHciC740dNz0","raw":"https://youtu.be/ITbwVFZYepc?si=_lAbRHciC740dNz0"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Edo Liberty discusses what real memory in LLMs requires beyond RAG - from scalable vector storage to reliable knowledge systems - and why storage, not compute, is becoming the key bottleneck for building dependable AI agents.","raw":"Edo Liberty discusses what real memory in LLMs requires beyond RAG - from scalable vector storage to reliable knowledge systems - and why storage, not compute, is becoming the key bottleneck for building dependable AI agents."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"3. Why AI Intelligence is Nothing Without Visual Memory | Shawn Shen on the Future of Embodied AI -> ","raw":"3. Why AI Intelligence is Nothing Without Visual Memory | Shawn Shen on the Future of Embodied AI -> "},{"type":"link","href":"https://youtu.be/3ccDi4ZczFg?si=SbJg487kwrkVXgUu","raw":"https://youtu.be/3ccDi4ZczFg?si=SbJg487kwrkVXgUu"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Shawn Shen argues AI needs a separate, hippocampus-like memory to move beyond chatbots, enabling long-term visual memory, object permanence, and on-device intelligence for robots, wearables, and the physical world","raw":"Shawn Shen argues AI needs a separate, hippocampus-like memory to move beyond chatbots, enabling long-term visual memory, object permanence, and on-device intelligence for robots, wearables, and the physical world"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"4. ","raw":"4. "},{"type":"resource","resource":{"type":"paper","id":"2504.15965"},"url":"https://huggingface.co/papers/2504.15965","raw":"https://huggingface.co/papers/2504.15965","label":"From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs (2504.15965)"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Links human memory types to LLM memory, introduces a taxonomy across object, form, and time, and identifies concrete limitations and future research directions","raw":"Links human memory types to LLM memory, introduces a taxonomy across object, form, and time, and identifies concrete limitations and future research directions"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"5. Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future Directions -> ","raw":"5. Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future Directions -> "},{"type":"link","href":"https://arxiv.org/abs/2505.00675v2","raw":"https://arxiv.org/abs/2505.00675v2"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Proposes a concrete taxonomy, core operations, and research directions to systematically organize and advance agent memory systems.","raw":"Proposes a concrete taxonomy, core operations, and research directions to systematically organize and advance agent memory systems."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Read further below ⬇️","raw":"Read further below ⬇️"},{"type":"new_line","raw":"\n"},{"type":"text","value":"If you like it, also subscribe to the Turing Post: ","raw":"If you like it, also subscribe to the Turing Post: "},{"type":"link","href":"https://www.turingpost.com/subscribe","raw":"https://www.turingpost.com/subscribe"}],"rawContent":"What we learned about memory in 2025: 8 comprehensive resources\n\nIf models forget everything, how can they be reliable? AI systems need to remember past interactions, update knowledge, stay consistent over time, and work beyond a single prompt. That's why many start to talk more about memory in AI. \nHere’s a useful set of studies and videos on where AI memory stands today:\n\n1. https://huggingface.co/papers/2512.13564\nA great survey that organizes agent memory research. It gives concrete taxonomies across memory form, function, and dynamics, summarizes benchmarks, frameworks, and emerging directions for building systematic agent memory systems\n\n2.When Will We Give AI True Memory? A conversation with Edo Liberty, CEO and founder @ Pinecone -> https://youtu.be/ITbwVFZYepc?si=_lAbRHciC740dNz0\nEdo Liberty discusses what real memory in LLMs requires beyond RAG - from scalable vector storage to reliable knowledge systems - and why storage, not compute, is becoming the key bottleneck for building dependable AI agents.\n\n3. Why AI Intelligence is Nothing Without Visual Memory | Shawn Shen on the Future of Embodied AI -> https://youtu.be/3ccDi4ZczFg?si=SbJg487kwrkVXgUu\nShawn Shen argues AI needs a separate, hippocampus-like memory to move beyond chatbots, enabling long-term visual memory, object permanence, and on-device intelligence for robots, wearables, and the physical world\n\n4. https://huggingface.co/papers/2504.15965\nLinks human memory types to LLM memory, introduces a taxonomy across object, form, and time, and identifies concrete limitations and future research directions\n\n5. Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future Directions -> https://arxiv.org/abs/2505.00675v2\nProposes a concrete taxonomy, core operations, and research directions to systematically organize and advance agent memory systems.\n\nRead further below ⬇️\nIf you like it, also subscribe to the Turing Post: https://www.turingpost.com/subscribe","author":{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},"attachments":[],"mentions":[],"reactions":[{"reaction":"👍","users":["John6666","panosdigital","victor","ashishk1331","ArchieCur","skew202"],"count":6}],"publishedAt":"2025-12-28T12:52:48.000Z","updatedAt":"2025-12-29T16:49:44.167Z","commentators":[{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},{"_id":"694fd87b6520314c5c9ac466","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/694fd87b6520314c5c9ac466/nJbLIW-61JNXQfthJnYLa.jpeg","fullname":"Mohamed Selim","name":"caesareg","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":5,"isUserFollowing":false}],"url":"/posts/Kseniase/637856294883643","totalUniqueImpressions":2474,"identifiedLanguage":{"language":"en","probability":0.8066526055335999},"numComments":3}},{"time":"2025-12-21T15:32:12.583Z","user":"Kseniase","userAvatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","type":"social-post-comment","socialPost":{"slug":"746586334382009","content":[{"type":"text","value":"From Prompt Engineering to Context Engineering: Main Design Patterns","raw":"From Prompt Engineering to Context Engineering: Main Design Patterns"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Earlier on, we relied on clever prompt wording, but now structured, complete context matters more than just magic phrasing. The next year is going to be a year of context engineering which expands beyond prompt engineering. The two complement each other: prompt engineering shapes how we ask, while context engineering shapes what the model knows, sees, and can do.","raw":"Earlier on, we relied on clever prompt wording, but now structured, complete context matters more than just magic phrasing. The next year is going to be a year of context engineering which expands beyond prompt engineering. The two complement each other: prompt engineering shapes how we ask, while context engineering shapes what the model knows, sees, and can do."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"To keep things clear, here are the main techniques and design patterns in both areas, with some useful resources for further exploration:","raw":"To keep things clear, here are the main techniques and design patterns in both areas, with some useful resources for further exploration:"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"▪️ 9 Prompt Engineering Techniques (configuring input text)","raw":"▪️ 9 Prompt Engineering Techniques (configuring input text)"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"1. Zero-shot prompting – giving a single instruction without examples. Relies entirely on pretrained knowledge.","raw":"1. Zero-shot prompting – giving a single instruction without examples. Relies entirely on pretrained knowledge."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"2. Few-shot prompting – adding input–output examples to encourage model to show the desired behavior. ⟶ ","raw":"2. Few-shot prompting – adding input–output examples to encourage model to show the desired behavior. ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2005.14165","raw":"https://arxiv.org/abs/2005.14165"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"3. Role prompting – assigning a persona or role (e.g. \"You are a senior researcher,\" \"Say it as a specialist in healthcare\") to shape style and reasoning. ⟶ ","raw":"3. Role prompting – assigning a persona or role (e.g. \"You are a senior researcher,\" \"Say it as a specialist in healthcare\") to shape style and reasoning. ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2403.02756","raw":"https://arxiv.org/abs/2403.02756"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"4. Instruction-based prompting – explicit constraints or guidance, like \"think step by step,\" \"use bullet points,\" \"answer in 10 words\"","raw":"4. Instruction-based prompting – explicit constraints or guidance, like \"think step by step,\" \"use bullet points,\" \"answer in 10 words\""},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"5. Chain-of-Thought (CoT) – encouraging intermediate reasoning traces to improve multi-step reasoning. It can be explicit (\"let’s think step by step\"), or implicit (demonstrated via examples). ⟶ ","raw":"5. Chain-of-Thought (CoT) – encouraging intermediate reasoning traces to improve multi-step reasoning. It can be explicit (\"let’s think step by step\"), or implicit (demonstrated via examples). ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2201.11903","raw":"https://arxiv.org/abs/2201.11903"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"6. Tree-of-Thought (ToT) – the model explores multiple reasoning paths in parallel, like branches of a tree, instead of following a single chain of thought. ⟶ ","raw":"6. Tree-of-Thought (ToT) – the model explores multiple reasoning paths in parallel, like branches of a tree, instead of following a single chain of thought. ⟶ "},{"type":"link","href":"https://arxiv.org/pdf/2203.11171","raw":"https://arxiv.org/pdf/2203.11171"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"7. Reasoning–action prompting (ReAct-style) – prompting the model to interleave reasoning steps with explicit actions and observations. It defines action slots and lets the model generate a sequence of \"Thought → Action → Observation\" steps. ⟶ ","raw":"7. Reasoning–action prompting (ReAct-style) – prompting the model to interleave reasoning steps with explicit actions and observations. It defines action slots and lets the model generate a sequence of \"Thought → Action → Observation\" steps. ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2210.03629","raw":"https://arxiv.org/abs/2210.03629"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Read further ⬇️","raw":"Read further ⬇️"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Also subscribe to Turing Post: ","raw":"Also subscribe to Turing Post: "},{"type":"link","href":"https://www.turingpost.com/subscribe","raw":"https://www.turingpost.com/subscribe"}],"rawContent":"From Prompt Engineering to Context Engineering: Main Design Patterns\n\nEarlier on, we relied on clever prompt wording, but now structured, complete context matters more than just magic phrasing. The next year is going to be a year of context engineering which expands beyond prompt engineering. The two complement each other: prompt engineering shapes how we ask, while context engineering shapes what the model knows, sees, and can do.\n\nTo keep things clear, here are the main techniques and design patterns in both areas, with some useful resources for further exploration:\n\n▪️ 9 Prompt Engineering Techniques (configuring input text)\n\n1. Zero-shot prompting – giving a single instruction without examples. Relies entirely on pretrained knowledge.\n\n2. Few-shot prompting – adding input–output examples to encourage model to show the desired behavior. ⟶ https://arxiv.org/abs/2005.14165\n\n3. Role prompting – assigning a persona or role (e.g. \"You are a senior researcher,\" \"Say it as a specialist in healthcare\") to shape style and reasoning. ⟶ https://arxiv.org/abs/2403.02756\n\n4. Instruction-based prompting – explicit constraints or guidance, like \"think step by step,\" \"use bullet points,\" \"answer in 10 words\"\n\n5. Chain-of-Thought (CoT) – encouraging intermediate reasoning traces to improve multi-step reasoning. It can be explicit (\"let’s think step by step\"), or implicit (demonstrated via examples). ⟶ https://arxiv.org/abs/2201.11903\n\n6. Tree-of-Thought (ToT) – the model explores multiple reasoning paths in parallel, like branches of a tree, instead of following a single chain of thought. ⟶ https://arxiv.org/pdf/2203.11171\n\n7. Reasoning–action prompting (ReAct-style) – prompting the model to interleave reasoning steps with explicit actions and observations. It defines action slots and lets the model generate a sequence of \"Thought → Action → Observation\" steps. ⟶ https://arxiv.org/abs/2210.03629\n\nRead further ⬇️\nAlso subscribe to Turing Post: https://www.turingpost.com/subscribe","author":{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},"attachments":[],"mentions":[],"reactions":[{"reaction":"❤️","users":["efecelik","John6666","Reubencf","sezginer","sasa2000","j14i"],"count":6}],"publishedAt":"2025-12-21T15:31:32.000Z","updatedAt":"2025-12-23T15:17:58.158Z","commentators":[{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},{"_id":"662da0e62d4c0e85daab7cc0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/662da0e62d4c0e85daab7cc0/Fx7WHD5t8San_GmjD_Dpl.jpeg","fullname":"efe","name":"efecelik","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":25,"isUserFollowing":false},{"_id":"6656f5950a10860d4c2f116e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6656f5950a10860d4c2f116e/iD_j4QgeOhmDoUxLSs9rP.jpeg","fullname":"Alex","name":"AlexPoto","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}],"url":"/posts/Kseniase/746586334382009","totalUniqueImpressions":3745,"identifiedLanguage":{"language":"en","probability":0.8534476161003113},"numComments":3},"comment":{"_id":"694812fcedae764cb1ae8813","raw":"8. Prompt templates – reusable prompt structures with placeholders for inputs.\n\n9. Retrieval-augmented prompting – injecting external information into the prompt. ⟶ https://arxiv.org/abs/2512.04106\n\nP.S. These are only the main prompt engineering techniques, but there are much more. Here's a good survey where you can find them: https://arxiv.org/pdf/2402.07927\n\n____\n\n▪️ Context Engineering (designing the information environment the model operates in)\n\n1. Retrieval-Augmented Generation (RAG) – dynamically injecting external knowledge retrieved from databases, search, or vector stores.\n\n2. Tool calling / function calling – enabling the model to use external tools. (APIs, calculators, code, search). They return extra info that the model needs to perform the task.\n\n3. Structured context – providing schemas, JSON, tables, or graphs instead of free-form text.\n\n4. System prompts / policies – persistent high-level instructions that govern behavior across interactions. ⟶ https://arxiv.org/abs/2212.08073\n\n5. Short-term memory – passing recent interaction history or intermediate state; summarizing information from the ongoing conversation. ⟶ https://arxiv.org/pdf/2512.13564\n\n6. Long-term memory – storing and retrieving user profiles, facts, or past decisions and conversations over time. ⟶ https://arxiv.org/abs/2503.08026\n\n7. Environment state – exposing the current world, task, or agent state (files, variables, observations).\n\n8. Multi-agent context – sharing state or messages between multiple LLM-based agents. ⟶ https://arxiv.org/abs/2505.21471","html":"

\n

1. Prompt templates – reusable prompt structures with placeholders for inputs.

   \n
\n

2. Retrieval-augmented prompting – injecting external information into the prompt. ⟶ https://arxiv.org/abs/2512.04106

   \n
\n

\n

P.S. These are only the main prompt engineering techniques, but there are much more. Here's a good survey where you can find them: https://arxiv.org/pdf/2402.07927

\n

---

\n

▪️ Context Engineering (designing the information environment the model operates in)

\n

\n

1. Retrieval-Augmented Generation (RAG) – dynamically injecting external knowledge retrieved from databases, search, or vector stores.

   \n
\n

2. Tool calling / function calling – enabling the model to use external tools. (APIs, calculators, code, search). They return extra info that the model needs to perform the task.

   \n
\n

3. Structured context – providing schemas, JSON, tables, or graphs instead of free-form text.

   \n
\n

4. System prompts / policies – persistent high-level instructions that govern behavior across interactions. ⟶ https://arxiv.org/abs/2212.08073

   \n
\n

5. Short-term memory – passing recent interaction history or intermediate state; summarizing information from the ongoing conversation. ⟶ https://arxiv.org/pdf/2512.13564

   \n
\n

6. Long-term memory – storing and retrieving user profiles, facts, or past decisions and conversations over time. ⟶ https://arxiv.org/abs/2503.08026

   \n
\n

7. Environment state – exposing the current world, task, or agent state (files, variables, observations).

   \n
\n

8. Multi-agent context – sharing state or messages between multiple LLM-based agents. ⟶ https://arxiv.org/abs/2505.21471

   \n
\n

\n"}}],"blogPosts":[{"_id":"685daf5f718dc84e7a6484f6","isUpvotedByUser":false,"numCoauthors":1,"publishedAt":"2025-06-26T21:19:16.531Z","slug":"codingreport","title":"What Coding Agent Wins? ","upvotes":9,"upvotes7d":0,"authorsData":[{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false}],"url":"/blog/Kseniase/codingreport","apiPrefix":"/api/blog/Kseniase/codingreport","canonical":false,"status":"published","isHFChangelog":false},{"_id":"681b5989894fcc0e9e93fc69","isUpvotedByUser":false,"numCoauthors":0,"publishedAt":"2025-05-07T14:22:44.006Z","slug":"a2a","title":"🦸🏻#17: What is A2A and why is it – still! – underappreciated?","upvotes":15,"upvotes7d":0,"authorsData":[{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false}],"url":"/blog/Kseniase/a2a","apiPrefix":"/api/blog/Kseniase/a2a","canonical":false,"status":"published","isHFChangelog":false}],"totalBlogPosts":42,"canReadDatabase":false,"canManageEntities":false,"canReadEntities":false,"canImpersonate":false,"canManageBilling":false,"canReadRepos":false,"canReadSpaces":false,"communityScore":1,"collections":[{"slug":"Kseniase/reinforcement-fine-tuning-67cdb9cfd524b2d8feca284d","title":"Reinforcement Fine Tuning","gating":false,"lastUpdated":"2025-03-09T15:54:55.796Z","owner":{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},"items":[{"_id":"67cdb9cfd524b2d8feca284e","position":0,"type":"paper","id":"2503.01785","title":"Visual-RFT: Visual Reinforcement Fine-Tuning","thumbnailUrl":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2503.01785.png","upvotes":86,"publishedAt":"2025-03-03T18:16:32.000Z","isUpvotedByUser":false}],"position":0,"theme":"purple","private":false,"shareUrl":"https://hf.co/collections/Kseniase/reinforcement-fine-tuning","upvotes":0,"isUpvotedByUser":false},{"slug":"Kseniase/mixture-of-experts-and-co-67cdb88ca26e6d3a70c04aaf","title":"Mixture-of-Experts and Co","gating":false,"lastUpdated":"2025-03-09T15:49:32.956Z","owner":{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},"items":[{"_id":"67cdb88ca26e6d3a70c04ab0","position":0,"type":"paper","id":"2503.02495","title":"Union of Experts: Adapting Hierarchical Routing to Equivalently\n Decomposed Transformer","thumbnailUrl":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2503.02495.png","upvotes":9,"publishedAt":"2025-03-04T11:01:25.000Z","isUpvotedByUser":false}],"position":1,"theme":"blue","private":false,"shareUrl":"https://hf.co/collections/Kseniase/mixture-of-experts-and-co","upvotes":0,"isUpvotedByUser":false},{"slug":"Kseniase/query-optimization-67991d09f704ff4ef3c25dd9","title":"Query Optimization","gating":false,"lastUpdated":"2025-01-28T19:04:47.033Z","owner":{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},"items":[{"_id":"67991d09f704ff4ef3c25dda","position":0,"type":"paper","id":"2401.06311","title":"Exploring the Best Practices of Query Expansion with Large Language\n Models","thumbnailUrl":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2401.06311.png","upvotes":1,"publishedAt":"2024-01-12T00:48:35.000Z","isUpvotedByUser":false},{"_id":"67991d87c9021c322608ea9c","position":1,"type":"paper","id":"2305.13954","title":"Robust Prompt Optimization for Large Language Models Against\n Distribution Shifts","thumbnailUrl":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2305.13954.png","upvotes":1,"publishedAt":"2023-05-23T11:30:43.000Z","isUpvotedByUser":false},{"_id":"67991d8db170998b0b6044dd","position":2,"type":"paper","id":"2407.12325","title":"Optimizing Query Generation for Enhanced Document Retrieval in RAG","thumbnailUrl":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2407.12325.png","upvotes":1,"publishedAt":"2024-07-17T05:50:32.000Z","isUpvotedByUser":false},{"_id":"679925f6ab65305cce9aa3f7","position":3,"type":"paper","id":"2408.00878","title":"Multi-Aspect Reviewed-Item Retrieval via LLM Query Decomposition and\n Aspect Fusion","thumbnailUrl":"https://cdn-thumbnails.huggingface.co/social-thumbnails/papers/2408.00878.png","upvotes":1,"publishedAt":"2024-08-01T19:04:10.000Z","isUpvotedByUser":false}],"position":2,"theme":"indigo","private":false,"shareUrl":"https://hf.co/collections/Kseniase/query-optimization","upvotes":0,"isUpvotedByUser":false}],"datasets":[],"models":[],"buckets":[],"numBuckets":0,"numberLikes":3,"papers":[],"posts":[{"slug":"637856294883643","content":[{"type":"text","value":"What we learned about memory in 2025: 8 comprehensive resources","raw":"What we learned about memory in 2025: 8 comprehensive resources"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"If models forget everything, how can they be reliable? AI systems need to remember past interactions, update knowledge, stay consistent over time, and work beyond a single prompt. That's why many start to talk more about memory in AI. ","raw":"If models forget everything, how can they be reliable? AI systems need to remember past interactions, update knowledge, stay consistent over time, and work beyond a single prompt. That's why many start to talk more about memory in AI. "},{"type":"new_line","raw":"\n"},{"type":"text","value":"Here’s a useful set of studies and videos on where AI memory stands today:","raw":"Here’s a useful set of studies and videos on where AI memory stands today:"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"1. ","raw":"1. "},{"type":"resource","resource":{"type":"paper","id":"2512.13564"},"url":"https://huggingface.co/papers/2512.13564","raw":"https://huggingface.co/papers/2512.13564","label":"Memory in the Age of AI Agents (2512.13564)"},{"type":"new_line","raw":"\n"},{"type":"text","value":"A great survey that organizes agent memory research. It gives concrete taxonomies across memory form, function, and dynamics, summarizes benchmarks, frameworks, and emerging directions for building systematic agent memory systems","raw":"A great survey that organizes agent memory research. It gives concrete taxonomies across memory form, function, and dynamics, summarizes benchmarks, frameworks, and emerging directions for building systematic agent memory systems"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"2.When Will We Give AI True Memory? A conversation with Edo Liberty, CEO and founder @ Pinecone -> ","raw":"2.When Will We Give AI True Memory? A conversation with Edo Liberty, CEO and founder @ Pinecone -> "},{"type":"link","href":"https://youtu.be/ITbwVFZYepc?si=_lAbRHciC740dNz0","raw":"https://youtu.be/ITbwVFZYepc?si=_lAbRHciC740dNz0"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Edo Liberty discusses what real memory in LLMs requires beyond RAG - from scalable vector storage to reliable knowledge systems - and why storage, not compute, is becoming the key bottleneck for building dependable AI agents.","raw":"Edo Liberty discusses what real memory in LLMs requires beyond RAG - from scalable vector storage to reliable knowledge systems - and why storage, not compute, is becoming the key bottleneck for building dependable AI agents."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"3. Why AI Intelligence is Nothing Without Visual Memory | Shawn Shen on the Future of Embodied AI -> ","raw":"3. Why AI Intelligence is Nothing Without Visual Memory | Shawn Shen on the Future of Embodied AI -> "},{"type":"link","href":"https://youtu.be/3ccDi4ZczFg?si=SbJg487kwrkVXgUu","raw":"https://youtu.be/3ccDi4ZczFg?si=SbJg487kwrkVXgUu"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Shawn Shen argues AI needs a separate, hippocampus-like memory to move beyond chatbots, enabling long-term visual memory, object permanence, and on-device intelligence for robots, wearables, and the physical world","raw":"Shawn Shen argues AI needs a separate, hippocampus-like memory to move beyond chatbots, enabling long-term visual memory, object permanence, and on-device intelligence for robots, wearables, and the physical world"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"4. ","raw":"4. "},{"type":"resource","resource":{"type":"paper","id":"2504.15965"},"url":"https://huggingface.co/papers/2504.15965","raw":"https://huggingface.co/papers/2504.15965","label":"From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs (2504.15965)"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Links human memory types to LLM memory, introduces a taxonomy across object, form, and time, and identifies concrete limitations and future research directions","raw":"Links human memory types to LLM memory, introduces a taxonomy across object, form, and time, and identifies concrete limitations and future research directions"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"5. Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future Directions -> ","raw":"5. Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future Directions -> "},{"type":"link","href":"https://arxiv.org/abs/2505.00675v2","raw":"https://arxiv.org/abs/2505.00675v2"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Proposes a concrete taxonomy, core operations, and research directions to systematically organize and advance agent memory systems.","raw":"Proposes a concrete taxonomy, core operations, and research directions to systematically organize and advance agent memory systems."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Read further below ⬇️","raw":"Read further below ⬇️"},{"type":"new_line","raw":"\n"},{"type":"text","value":"If you like it, also subscribe to the Turing Post: ","raw":"If you like it, also subscribe to the Turing Post: "},{"type":"link","href":"https://www.turingpost.com/subscribe","raw":"https://www.turingpost.com/subscribe"}],"rawContent":"What we learned about memory in 2025: 8 comprehensive resources\n\nIf models forget everything, how can they be reliable? AI systems need to remember past interactions, update knowledge, stay consistent over time, and work beyond a single prompt. That's why many start to talk more about memory in AI. \nHere’s a useful set of studies and videos on where AI memory stands today:\n\n1. https://huggingface.co/papers/2512.13564\nA great survey that organizes agent memory research. It gives concrete taxonomies across memory form, function, and dynamics, summarizes benchmarks, frameworks, and emerging directions for building systematic agent memory systems\n\n2.When Will We Give AI True Memory? A conversation with Edo Liberty, CEO and founder @ Pinecone -> https://youtu.be/ITbwVFZYepc?si=_lAbRHciC740dNz0\nEdo Liberty discusses what real memory in LLMs requires beyond RAG - from scalable vector storage to reliable knowledge systems - and why storage, not compute, is becoming the key bottleneck for building dependable AI agents.\n\n3. Why AI Intelligence is Nothing Without Visual Memory | Shawn Shen on the Future of Embodied AI -> https://youtu.be/3ccDi4ZczFg?si=SbJg487kwrkVXgUu\nShawn Shen argues AI needs a separate, hippocampus-like memory to move beyond chatbots, enabling long-term visual memory, object permanence, and on-device intelligence for robots, wearables, and the physical world\n\n4. https://huggingface.co/papers/2504.15965\nLinks human memory types to LLM memory, introduces a taxonomy across object, form, and time, and identifies concrete limitations and future research directions\n\n5. Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future Directions -> https://arxiv.org/abs/2505.00675v2\nProposes a concrete taxonomy, core operations, and research directions to systematically organize and advance agent memory systems.\n\nRead further below ⬇️\nIf you like it, also subscribe to the Turing Post: https://www.turingpost.com/subscribe","author":{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},"attachments":[],"mentions":[],"reactions":[{"reaction":"👍","users":["John6666","panosdigital","victor","ashishk1331","ArchieCur","skew202"],"count":6}],"publishedAt":"2025-12-28T12:52:48.000Z","updatedAt":"2025-12-29T16:49:44.167Z","commentators":[{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},{"_id":"694fd87b6520314c5c9ac466","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/694fd87b6520314c5c9ac466/nJbLIW-61JNXQfthJnYLa.jpeg","fullname":"Mohamed Selim","name":"caesareg","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":5,"isUserFollowing":false}],"url":"/posts/Kseniase/637856294883643","totalUniqueImpressions":2474,"identifiedLanguage":{"language":"en","probability":0.8066526055335999},"numComments":3},{"slug":"746586334382009","content":[{"type":"text","value":"From Prompt Engineering to Context Engineering: Main Design Patterns","raw":"From Prompt Engineering to Context Engineering: Main Design Patterns"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Earlier on, we relied on clever prompt wording, but now structured, complete context matters more than just magic phrasing. The next year is going to be a year of context engineering which expands beyond prompt engineering. The two complement each other: prompt engineering shapes how we ask, while context engineering shapes what the model knows, sees, and can do.","raw":"Earlier on, we relied on clever prompt wording, but now structured, complete context matters more than just magic phrasing. The next year is going to be a year of context engineering which expands beyond prompt engineering. The two complement each other: prompt engineering shapes how we ask, while context engineering shapes what the model knows, sees, and can do."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"To keep things clear, here are the main techniques and design patterns in both areas, with some useful resources for further exploration:","raw":"To keep things clear, here are the main techniques and design patterns in both areas, with some useful resources for further exploration:"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"▪️ 9 Prompt Engineering Techniques (configuring input text)","raw":"▪️ 9 Prompt Engineering Techniques (configuring input text)"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"1. Zero-shot prompting – giving a single instruction without examples. Relies entirely on pretrained knowledge.","raw":"1. Zero-shot prompting – giving a single instruction without examples. Relies entirely on pretrained knowledge."},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"2. Few-shot prompting – adding input–output examples to encourage model to show the desired behavior. ⟶ ","raw":"2. Few-shot prompting – adding input–output examples to encourage model to show the desired behavior. ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2005.14165","raw":"https://arxiv.org/abs/2005.14165"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"3. Role prompting – assigning a persona or role (e.g. \"You are a senior researcher,\" \"Say it as a specialist in healthcare\") to shape style and reasoning. ⟶ ","raw":"3. Role prompting – assigning a persona or role (e.g. \"You are a senior researcher,\" \"Say it as a specialist in healthcare\") to shape style and reasoning. ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2403.02756","raw":"https://arxiv.org/abs/2403.02756"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"4. Instruction-based prompting – explicit constraints or guidance, like \"think step by step,\" \"use bullet points,\" \"answer in 10 words\"","raw":"4. Instruction-based prompting – explicit constraints or guidance, like \"think step by step,\" \"use bullet points,\" \"answer in 10 words\""},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"5. Chain-of-Thought (CoT) – encouraging intermediate reasoning traces to improve multi-step reasoning. It can be explicit (\"let’s think step by step\"), or implicit (demonstrated via examples). ⟶ ","raw":"5. Chain-of-Thought (CoT) – encouraging intermediate reasoning traces to improve multi-step reasoning. It can be explicit (\"let’s think step by step\"), or implicit (demonstrated via examples). ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2201.11903","raw":"https://arxiv.org/abs/2201.11903"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"6. Tree-of-Thought (ToT) – the model explores multiple reasoning paths in parallel, like branches of a tree, instead of following a single chain of thought. ⟶ ","raw":"6. Tree-of-Thought (ToT) – the model explores multiple reasoning paths in parallel, like branches of a tree, instead of following a single chain of thought. ⟶ "},{"type":"link","href":"https://arxiv.org/pdf/2203.11171","raw":"https://arxiv.org/pdf/2203.11171"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"7. Reasoning–action prompting (ReAct-style) – prompting the model to interleave reasoning steps with explicit actions and observations. It defines action slots and lets the model generate a sequence of \"Thought → Action → Observation\" steps. ⟶ ","raw":"7. Reasoning–action prompting (ReAct-style) – prompting the model to interleave reasoning steps with explicit actions and observations. It defines action slots and lets the model generate a sequence of \"Thought → Action → Observation\" steps. ⟶ "},{"type":"link","href":"https://arxiv.org/abs/2210.03629","raw":"https://arxiv.org/abs/2210.03629"},{"type":"new_line","raw":"\n"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Read further ⬇️","raw":"Read further ⬇️"},{"type":"new_line","raw":"\n"},{"type":"text","value":"Also subscribe to Turing Post: ","raw":"Also subscribe to Turing Post: "},{"type":"link","href":"https://www.turingpost.com/subscribe","raw":"https://www.turingpost.com/subscribe"}],"rawContent":"From Prompt Engineering to Context Engineering: Main Design Patterns\n\nEarlier on, we relied on clever prompt wording, but now structured, complete context matters more than just magic phrasing. The next year is going to be a year of context engineering which expands beyond prompt engineering. The two complement each other: prompt engineering shapes how we ask, while context engineering shapes what the model knows, sees, and can do.\n\nTo keep things clear, here are the main techniques and design patterns in both areas, with some useful resources for further exploration:\n\n▪️ 9 Prompt Engineering Techniques (configuring input text)\n\n1. Zero-shot prompting – giving a single instruction without examples. Relies entirely on pretrained knowledge.\n\n2. Few-shot prompting – adding input–output examples to encourage model to show the desired behavior. ⟶ https://arxiv.org/abs/2005.14165\n\n3. Role prompting – assigning a persona or role (e.g. \"You are a senior researcher,\" \"Say it as a specialist in healthcare\") to shape style and reasoning. ⟶ https://arxiv.org/abs/2403.02756\n\n4. Instruction-based prompting – explicit constraints or guidance, like \"think step by step,\" \"use bullet points,\" \"answer in 10 words\"\n\n5. Chain-of-Thought (CoT) – encouraging intermediate reasoning traces to improve multi-step reasoning. It can be explicit (\"let’s think step by step\"), or implicit (demonstrated via examples). ⟶ https://arxiv.org/abs/2201.11903\n\n6. Tree-of-Thought (ToT) – the model explores multiple reasoning paths in parallel, like branches of a tree, instead of following a single chain of thought. ⟶ https://arxiv.org/pdf/2203.11171\n\n7. Reasoning–action prompting (ReAct-style) – prompting the model to interleave reasoning steps with explicit actions and observations. It defines action slots and lets the model generate a sequence of \"Thought → Action → Observation\" steps. ⟶ https://arxiv.org/abs/2210.03629\n\nRead further ⬇️\nAlso subscribe to Turing Post: https://www.turingpost.com/subscribe","author":{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},"attachments":[],"mentions":[],"reactions":[{"reaction":"❤️","users":["efecelik","John6666","Reubencf","sezginer","sasa2000","j14i"],"count":6}],"publishedAt":"2025-12-21T15:31:32.000Z","updatedAt":"2025-12-23T15:17:58.158Z","commentators":[{"_id":"64838b28c235ef76b63e4999","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","fullname":"Ksenia Se","name":"Kseniase","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1337,"isUserFollowing":false},{"_id":"662da0e62d4c0e85daab7cc0","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/662da0e62d4c0e85daab7cc0/Fx7WHD5t8San_GmjD_Dpl.jpeg","fullname":"efe","name":"efecelik","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":25,"isUserFollowing":false},{"_id":"6656f5950a10860d4c2f116e","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6656f5950a10860d4c2f116e/iD_j4QgeOhmDoUxLSs9rP.jpeg","fullname":"Alex","name":"AlexPoto","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}],"url":"/posts/Kseniase/746586334382009","totalUniqueImpressions":3745,"identifiedLanguage":{"language":"en","probability":0.8534476161003113},"numComments":3}],"totalPosts":53,"spaces":[],"u":{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/ZhQCYoU3vps71Ag7Jezj6.jpeg","isPro":false,"fullname":"Ksenia Se","user":"Kseniase","orgs":[{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64838b28c235ef76b63e4999/P43kYfo2cFwf_fM5oPaWT.png","fullname":"Turing Post","name":"TuringPost","userRole":"admin","type":"org","isHf":false,"plan":"team"},{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63691c3eda9b693c2730b2a2/WoOIHdJahrAnLo1wpyVjc.png","fullname":"Journalists on Hugging Face","name":"JournalistsonHF","userRole":"contributor","type":"org","isHf":false,"details":"Democratizing access to useful AI tools and resources for journalists"},{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/5f17f0a0925b9863e28ad517/nxmdd6m86cxu55UZBlQeg.jpeg","fullname":"Social Post Explorers","name":"social-post-explorers","userRole":"read","type":"org","isHf":false},{"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6340651b388c3fa40f9a5bc0/j6Vb_hutYuKRcQgMaDTAt.png","fullname":"Hugging Face Discord Community","name":"discord-community","userRole":"read","type":"org","isHf":false,"details":"Collaborating towards Good ML!"},{"avatarUrl":"https://www.gravatar.com/avatar/2cc245ecb4e7dd7f0526018fce64190c?d=retro&size=100","fullname":"Sandbox","name":"projects-sandbox","userRole":"read","type":"org","isHf":false}],"signup":{"homepage":"https://www.turingpost.com/","twitter":"Kseniase_","details":"","github":"","bluesky":"kseniase.bsky.social","linkedin":"ksenia-se"},"isHf":false,"isMod":false,"type":"user","theme":"light"},"upvotes":100,"numFollowers":1337,"numFollowingUsers":115,"numFollowingOrgs":11,"numModels":0,"numDatasets":0,"numSpaces":0,"isFollowing":false,"isFollower":false,"sampleFollowers":[{"user":"lH1234","fullname":"lynn h","type":"user","_id":"6459db665e44a749d5a693bd","isPro":false,"avatarUrl":"/avatars/3b69a9f73b8cab5ab0140b96775ab816.svg"},{"user":"yjernite","fullname":"Yacine Jernite","type":"user","_id":"5ee3a7cd2a3eae3cbdad1305","isPro":true,"avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1594144055859-5ee3a7cd2a3eae3cbdad1305.jpeg"},{"user":"olefevre09","fullname":"Lefevre Olivier","type":"user","_id":"644a1a744d8c0015680214ca","isPro":false,"avatarUrl":"/avatars/59181f55f5a4cba030ba23b1a371be54.svg"},{"user":"syedasad","fullname":"Syed Asad Quadri","type":"user","_id":"65602a9a5b627f441ad9537c","isPro":false,"avatarUrl":"/avatars/396fc0de44b810f47ccfeb8408514d0f.svg"}],"isWatching":false,"isIgnored":false,"acceptLanguages":["*"],"filters":{},"currentRepoPage":0}">