Unverified50% confidenceFactExact time
VIDRAFT's same NPU solution already covers model lifecycle stages including pretraining from scratch, LoRA fine-tuning, inference acceleration (also validated on general open-source models), and quantized model execution and acceleration.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/14/2026
First Seen
Valid until: 1/12/2027
Sources
Related Entities
Related Claims
UnverifiedVIDRAFT completed full backpropagation on an inference-specific NPU, pretrained a language model from scratch, and released it as VKORE-46M-NPU.72% similarUnverified零样本能力指模型在未经针对该任务专项训练的情况下直接完成推理，是衡量模型是否内化物理化学规律的试金石72% similarUnverifiedGTF mixes true observations and the model's own predictions in proportion via a tunable mixing coefficient before feeding the result into the next step, biasing toward true data in the early stages of training and gradually increasing the proportion of the model's self-predictions as training progresses.71% similarUnverifiedThe demo setup uses open-source models from Nous Research by default and is configured with fallback models via OpenRouter (such as Tencent's Hunyuan model), automatically switching when the primary model triggers rate limits.70% similarUnverifiedUsing large-scale pre-trained models (such as Whisper and HuBERT) as a foundation for fine-tuning on specific tasks allows small datasets to be reframed as few-shot fine-tuning scenarios.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1026572API
curl https://kongchang.com/api/v1/knowledge/claims/1026572MCP
get_claim(id=1026572)