Unverified50% confidenceBenchmarkExact time
The LoRA+GRPO recipe summarized in the Latent Space AI news digest: trained with verl, using vLLM for rollout, running 150 steps on 16 GB300s, raising AIME-2024 mean@8 from 0.56 to 0.85.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/16/2026
First Seen
Valid until: 1/14/2027
Sources
Related Entities
Related Claims
UnverifiedThe RL-Kernel team ran a 200-step Qwen3-8B GRPO training task on 8 AMD MI300X cards, with results showing zero logprob mismatch between the Megatron training side and the vLLM rollout side.71% similarUnverifiedGrok 4.5 is based on the all-new V9 base model, with parameters reaching 1.5 trillion—three times those of the previous-generation V8—and was trained using tens of thousands of NVIDIA GB300 GPUs.67% similarUnverifiedGB200 NVL72是面向超大规模AI训练和推理场景的旗舰级计算平台61% similarUnverifiedThe SupraLabs team has released a lightweight text-to-image model, Supra2-IMG, with 100M parameters, trained from scratch using a DiT architecture, with an output resolution of 256×256, and it is fully open-source.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1044470API
curl https://kongchang.com/api/v1/knowledge/claims/1044470MCP
get_claim(id=1044470)