shenzhen
Joined February 2007
Datou retweeted
kale-img-1 fine-tuning results 1. @ostrisai / 8.10 / moss 2. @AmandineFlachs / 7.61 / moss 3. @suhaaspk / 7.20 / moss 4. @Assu_Bhadhu / 6.90 / moss 5. @max_pe2002 / 5.35 / hdo1 6. @hasnain__dev / 3.51 / sdxl (same prompt, seed, resolution)
paying $5k to whoever fine-tunes kale-img-1, an img model for kids dm me i'll send u the dataset u pick a base model we can license commercially u comment with ur result here ur result gets judged on: - recreating 5 imgs from the dataset almost 1:1 - making 5 new imgs in our style
4
2
2
20
1,122
5 MiniMax H3 speedups on one RTX 5090 Stock 20-step: 129.0 s larryvrh Turbo: 61.4 s (2.1x) LightX2V Turbo: 55.8 s (2.3x) FastH3 V2 +VSA: 44.7 s (2.9x) DMAD 4-step: 31.7 s (4.1x) Same diner clip, 5 s with sound, 1344x768. Settings and quality in the reply.
15
18
1
254
20,890
要对动森下手了
もうこれを遊ぶしかない、HoYoverseが贈る、キミの星をつくるスローライフゲーム。

61
grok这个动画对原图特征保留的真不错
This is just beautiful 😍
574
有点像内化了jev的快速判断能力
An agent needs to reason before acting. It generates a chain of thought token by token — slow, and small models struggle to produce coherent free-form reasoning. What if, instead of writing an essay from scratch every time, the model chose from a predefined menu of reasoning strategies — like answering a multiple-choice question? Apple and Carnegie Mellon University present SSR (Selection-based Structured Reasoning, ArXiv 2610.01892). SSR's core is a reasoning library of six reusable options: use image search to identify uncertain entities; use text search for specific facts; crop or zoom to inspect image details; answer based on collected evidence; read the answer directly from the image; or proceed when evidence is sufficient. Rather than generating reasoning from scratch, the model selects the appropriate option — and Parallel Reasoning Decoding evaluates candidates in parallel, which is the direct source of its speedup. On seven multimodal search benchmarks, SSR reduces per-round reasoning latency by over 90% while maintaining task success rates comparable to free-form thinking. In one example, SSR selects "image entity recognition" in just 32.66 milliseconds and immediately triggers an image search, where free-form generation would spend far longer composing the same reasoning. Paper, project page, and code are open. Result: SSR replaces token-by-token reasoning generation with selection from a 6-option reasoning library plus parallel decoding — >90% per-round latency reduction on 7 multimodal search benchmarks at comparable success rates — from Apple and CMU, with open paper, project, and code. SSR: Selection-based Structured Reasoning for Efficient Agent Inference Paper: arxiv.org/abs/2610.01892 Project: zfy0314.github.io/ssr-webpag… Our report: jiqizhixin.com/articles/2026… 📬 #PapersAccepted by Jiqizhixin
570
马斯克感觉是玩明白流量了, 引导语言不通的人吵架。
8
5
4
54
11,271
《过主线剧情的时候一定要穿的像个人》
1
9
2,136
猜猜这是干什么的?答案在评论区
1
1,073
人类有个著名的技能叫持久性狩猎
Replying to @StarKnight
其实在人类历史上的绝大部分时间,绝大部分人群,都没有经历过“每天一小时以上通勤”这回事。
2
864
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model. 🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal. 🔹 Full-project coding: games in Unreal, Unity & Godot. 🔹 Excels at parametric 3D modeling in Blender & CAD. 🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%. 🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks. Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. Open weights by Nov 4, 2026. 🔗 API: platform.kimi.ai 🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
1,239
3,226
2,351
30,092
3,992,528
特吹怎么看?
$TSLA crashed, erupted in flames 💀 Driver trapped burned alive 💀 Where were Tesla advanced safety features, automatic emergency braking (AEB), automatic collision avoidance? 💀 For every "Tesla FSD saved my family" claim ... there are more REAL incidents like this 👇
4
5
1,241
Datou retweeted
I’ve developed TARA — Temporally Aligned Reference Attention, an experimental method for MiniMax H3. Research, implementation and video examples: github.com/alisson-anjos/ai-… Native video references can capture an action but drift out of sync with the source. TARA aligns the reference and output timelines inside attention, helping preserve performance and timing while using native reference conditioning, without Add Guide. I’ve also developed optional precompiled kernels for Comfy Kitchen and sol-attn, with TARA’s temporal attention bias fused directly into the attention computation. This avoids the slower per-frame compatibility path. BFSNodes automatically downloads the matching binary on supported systems, compatibility and performance across GPUs are still being validated. Early tests look promising. Interestingly, my guide-latent-trained LoRA currently seems to work better with TARA than the unfinished TARA-trained checkpoint. This isn’t a controlled comparison; I’m training a fresh LoRA from scratch to investigate further. There’s a TARA node in BFSNodes for native ComfyUI workflows. If you’d like to try it, feedback, settings and short comparisons are very welcome, especially from different GPUs!
8
14
1
148
6,608
卧槽,新商机:给 AI 配手机 把手机插在机架上 摄像头对着屏幕 机械臂握着触控笔操控手机 完美帮 Agent 绕开 App 风控 国外已经有人开始做了
610
Datou retweeted
1984년 매킨토시로 변한 아이폰 듀오 애플의 첫 폴더블폰 〈iPhone Duo〉를 반쯤 접으면 1984년 매킨토시를 재현한 작은 노트북으로 변신합니다. 새롭게 공개된 〈Tiny Computers〉는 매킨토시부터 윈도우 95·XP까지 고전 컴퓨터 10종을 체험할 수 있는 앱입니다. 키보드 입력은 물론 그림 그리기와 간단한 게임도 즐길 수 있으며, 일반 아이폰과 아이패드에서도 이용할 수 있습니다. 🎥 idanbo
152
1,889
417
11,717
1,983,759
【MiniMax H3のキャラクター表現を改善するオリジナルLoRA学習とパイプライン検証】 動画生成モデル「MiniMax H3」でオリジナルキャラクターの品質を向上させるため、LoRA学習と各種パイプラインを比較検証した知見が共有されています! RefMod(latent references)単体では、映像全体が過度に磨きすぎた不自然な印象になりがちです。そこで、静止画8枚、音声、会話クリップ14本を組み合わせてキャラクターLoRAを構築することで、t2v本来の自然な質感を維持したままキャラクターの一貫性を担保できるのが大きなポイントですね。 学習環境にはModalのH200 GPUを使用し、約41分で完了しています!推論環境はRTX 3090で、LoRAの適用強度は0.8がスイートスポットとのことです。1.0にするとファンデーションを厚塗りしたような見た目になってしまいます。 リファレンス画像がライティングを強く引き継ぐ点や、トレーニングプレビューを過信してはいけないといった実践的な教訓も参考になります! #MiniMaxH3 #StableDiffusion
5
36
1
395
20,800
MiniMax H3 のキャラ入れ替え LoRA (Character Swap v1.1) を試しました 結果は動画を見ての通り、 元の動画の動きを忠実にコピーしつつ、 キャラクタの一貫性も維持できてます 以前に LoRA なし V2V で何とかしようとして 苦労したのが嘘のよう ワークフローなどはリプ欄に 羽根付きキャラは、途中で一瞬羽根が消えたりしてますが、元動画との造形(体形)が違うので、これは仕方ない気はします むしろ、造形が違うキャラでも入れ替えが出来ているのは、凄いんじゃないでしょうか
Astra を使っても MiniMax H3 の V2V が 失敗する原因をいろいろ探ってましたが、 どうやら 「元動画の動きが速すぎる」 と失敗することが多いようです それで、元画像の速度を 0.8 倍とかにして V2V (ref2va) で生成し、その後に 1.25 倍すると 結構うまくいくことが分かりました これを Astra 君に伝えて、 「そうやって生成して!」とお願いして できたのが下の動画です 比較で Wan3.0/SeeDance 2.5 の生成動画も つけました この動画、通常の速度で MiniMax H3 で V2V しても全然うまくいきませんでした 0.8 倍にしたら(この動画は)いけましたが、 動画によってはもっと遅くする必要が あるかもしれません xitter.cf/IsekaiBardGuild/status…
5
35
5
420
25,679
🔥炸了!国产AI 3D模型Hi3D V3.0正式发布,细节反超GPT-6 Astra!V1.0同步开源! 同一张人物图对比:GPT-6 Astra只能搭出基本形体,Hi3D Agent+V3.0在轮廓、体积和局部细节上明显更贴近原图,直接拉开差距。 核心亮点: 🔹业界首个可商用2048³超高精度国产自研模型,纹理最高8K 🔹V1.0正式开源,基于Sparc3D,训练数据扩至约200万,切片成功率90%,已进生产流 🔹Agent闭环:意图理解→生成→持续修改→工程交付,支持MCP,可直接进Blender/游戏引擎 开源地址:hi3d.ai/hi3d-1-0-open-source
GitHub:github.com/hi3d-official/Hi3…
官网:hi3d.ai/ #科技 #国产AI #Hi3D #3D生成 #Agent #开源 #GPT6Astra
20
43
1
240
16,826