BAAI/OPI-Struc
Updated • 353 • 1
AI & ML interests
None defined yet.
Recent Activity
View all activity
Papers
ExoActor: Exocentric Video Generation as Generalizable Interactive Humanoid Control
AutoResearchBench: Benchmarking AI Agents on Complex Scientific Literature Discovery
01Matrix
updated a
dataset about 23 hours ago
wwen1997
authored 2
papers 3 days ago
Paranioar
authored 3
papers 7 days ago
DynamicVLA: A Vision-Language-Action Model for Dynamic Object Manipulation
Paper • 2601.22153 • Published • 75
VISTA-Bench: Do Vision-Language Models Really Understand Visualized Text as Well as Pure Text?
Paper • 2602.04802 • Published • 2
SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecture
Paper • 2605.12500 • Published • 184
A-PolarBear
updated a
model 8 days ago
Paranioar
submitted a
paper to Daily Papers 8 days ago
01Matrix
published a
dataset 10 days ago
ldwang
updated a
model 13 days ago
A-PolarBear
updated a
collection 16 days ago
A-PolarBear
published a
model 16 days ago
Bitterdhg
authored a
paper 17 days ago
tellarin
authored a
paper 17 days ago
Sri-Vigneshwar-DJ
posted an update 18 days ago
Post
124
We ran Feather DB v0.8.0 on LongMemEval (ICLR 2025) — 500 questions across real multi-session conversations, up to 115K tokens each.
**Score: 0.693** · GPT-4o full-context baseline: 0.640
Full 500-question run with Gemini-Flash: **$2.40**
Per-axis breakdown:
→ Info-extraction: **0.942**
→ Knowledge-update: **0.714**
→ Multi-session: **0.606**
→ Temporal: **0.477** ← the hard one, Phase 9 addresses this
Architecture: Hybrid BM25+dense · adaptive temporal decay · embedded (no server) · p50 = 0.19ms · MIT
Raw results + audit JSONs: Hawky-ai/longmemeval-results
We ran Feather DB v0.8.0 on LongMemEval (ICLR 2025) — 500 questions across real multi-session conversations, up to 115K tokens each.
**Score: 0.693** · GPT-4o full-context baseline: 0.640
Full 500-question run with Gemini-Flash: **$2.40**
Per-axis breakdown:
→ Info-extraction: **0.942**
→ Knowledge-update: **0.714**
→ Multi-session: **0.606**
→ Temporal: **0.477** ← the hard one, Phase 9 addresses this
Architecture: Hybrid BM25+dense · adaptive temporal decay · embedded (no server) · p50 = 0.19ms · MIT
pip install feather-dbRaw results + audit JSONs: Hawky-ai/longmemeval-results
tellarin
submitted a
paper to Daily Papers 20 days ago
Post
129
🚀 Run DeepSeek V4 on more AI GPUs with FlagOS
DeepSeek V4 just dropped with huge specs: 1.6T params, 1M context, MIT license.
But there’s a catch: the official weights use FP4+FP8 mixed precision, which mainly targets NVIDIA Blackwell / B200-class GPUs.
So we built DeepSeek-V4-FlagOS.
On Day 0, the FlagOS community completed multi-chip adaptation across 8 AI hardware platforms:
✅ NVIDIA H100/H20 — FP8/BF16
✅ Huawei Ascend — BF16
✅ Hygon DCU — BF16
✅ MetaX GPU — BF16
✅ Moore Threads MTT S5000 — FP8
✅ Kunlunxin XPU — BF16
✅ T-Head/Alibaba Zhenwu — BF16
✅ Iluvatar GPU — BF16
🔧 What makes it work?
1️⃣ FlagGems operator replacement
DeepSeek V4 operators — MoE routing, Attention, RMSNorm and more — are reimplemented with Triton, reducing dependency on CUDA-specific libraries.
New V4 operators include:
Act Quant, hc_split_sinkhorn, FP8 MatMul, Sparse Attention, Hadamard Transform.
2️⃣ Flexible tensor parallelism
DeepSeek V4 uses o_groups=8, which can limit TP.
We added an independent communication group for o-groups, while allowing the rest of the model to scale to higher TP, enabling deployment on 32GB/64GB cards.
3️⃣ FP4 → BF16 conversion
For hardware without native FP4, we provide ready-to-use BF16 conversion and pre-converted model releases.
📦 Pre-converted models are available on Hugging Face:
V4-Pro:
FlagRelease/DeepSeek-V4-Pro-nvidia-FlagOS
FlagRelease/DeepSeek-V4-Pro-metax-FlagOS
FlagRelease/DeepSeek-V4-Pro-mthreads-FlagOS
FlagRelease/DeepSeek-V4-Pro-hygon-FlagOS
FlagRelease/DeepSeek-V4-Pro-ascend-FlagOS
V4-Flash:
FlagRelease/DeepSeek-V4-Flash-nvidia-FlagOS
FlagRelease/DeepSeek-V4-Flash-zhenwu-FlagOS
FlagRelease/DeepSeek-V4-Flash-kunlunxin-FlagOS
FlagRelease/DeepSeek-V4-Flash-iluvatar-FlagOS
⚡ Performance on NVIDIA H20, V4-Flash FP8:
FlagGems C++ Wrapper + Triton: 70.7 tok/s
DeepSeek TileLang: 62.99 tok/s
That’s 12.24% faster.
👉 Try it here:
https://github.com/flagos-ai/DeepSeek-V4-FlagOS
Open models should run on open infrastructure
DeepSeek V4 just dropped with huge specs: 1.6T params, 1M context, MIT license.
But there’s a catch: the official weights use FP4+FP8 mixed precision, which mainly targets NVIDIA Blackwell / B200-class GPUs.
So we built DeepSeek-V4-FlagOS.
On Day 0, the FlagOS community completed multi-chip adaptation across 8 AI hardware platforms:
✅ NVIDIA H100/H20 — FP8/BF16
✅ Huawei Ascend — BF16
✅ Hygon DCU — BF16
✅ MetaX GPU — BF16
✅ Moore Threads MTT S5000 — FP8
✅ Kunlunxin XPU — BF16
✅ T-Head/Alibaba Zhenwu — BF16
✅ Iluvatar GPU — BF16
🔧 What makes it work?
1️⃣ FlagGems operator replacement
DeepSeek V4 operators — MoE routing, Attention, RMSNorm and more — are reimplemented with Triton, reducing dependency on CUDA-specific libraries.
New V4 operators include:
Act Quant, hc_split_sinkhorn, FP8 MatMul, Sparse Attention, Hadamard Transform.
2️⃣ Flexible tensor parallelism
DeepSeek V4 uses o_groups=8, which can limit TP.
We added an independent communication group for o-groups, while allowing the rest of the model to scale to higher TP, enabling deployment on 32GB/64GB cards.
3️⃣ FP4 → BF16 conversion
For hardware without native FP4, we provide ready-to-use BF16 conversion and pre-converted model releases.
📦 Pre-converted models are available on Hugging Face:
V4-Pro:
FlagRelease/DeepSeek-V4-Pro-nvidia-FlagOS
FlagRelease/DeepSeek-V4-Pro-metax-FlagOS
FlagRelease/DeepSeek-V4-Pro-mthreads-FlagOS
FlagRelease/DeepSeek-V4-Pro-hygon-FlagOS
FlagRelease/DeepSeek-V4-Pro-ascend-FlagOS
V4-Flash:
FlagRelease/DeepSeek-V4-Flash-nvidia-FlagOS
FlagRelease/DeepSeek-V4-Flash-zhenwu-FlagOS
FlagRelease/DeepSeek-V4-Flash-kunlunxin-FlagOS
FlagRelease/DeepSeek-V4-Flash-iluvatar-FlagOS
⚡ Performance on NVIDIA H20, V4-Flash FP8:
FlagGems C++ Wrapper + Triton: 70.7 tok/s
DeepSeek TileLang: 62.99 tok/s
That’s 12.24% faster.
👉 Try it here:
https://github.com/flagos-ai/DeepSeek-V4-FlagOS
Open models should run on open infrastructure
PhyscalX
updated a
collection 29 days ago
PhyscalX
authored a
paper 29 days ago
Alon77777
authored a
paper about 1 month ago