Unverified50% confidenceBenchmarkExact time
Mistral says the 209M version of Timestral ranks first among zero-shot pretrained models on GIFT-Eval/TIME/fev-bench.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/17/2026
First Seen
Valid until: 10/31/2026
Sources
Related Entities
Related Claims
UnverifiedMany downstream fine-tuned versions and quantized versions do not separately document the training data cutoff date.61% similarUnverifiedTabPFN v2 by default refuses more than approximately 1,000 rows of training data; setting ignore_pretraining_limits=True can push this to 10,000 rows, but a single fit takes 30–90 minutes instead of a few seconds.61% similarUnverifiedOn LLaDA-2.1-mini, ORL requires only 4.2M self-generated training tokens and 36 H100 GPU hours of post-training to obtain up to a 3x throughput improvement with comparable quality.60% similarUnverifiedTraining uses batch size 256, 400,000 steps, EMA 0.9999, linear learning rate decay over the final quarter, and torch.compile achieves a 2.4× speedup over eager mode.60% similarUnverifiedThe 20.6-second generation time does not include training cost; training used 16 V100 GPUs.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1052861API
curl https://kongchang.com/api/v1/knowledge/claims/1052861MCP
get_claim(id=1052861)