Unverified50% confidenceFactExact time
DeepSeek and Huawei jointly optimized the Ascend 950 large-scale cluster solution, performing in-depth tuning for multi-card training and large model inference.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/1/2026
First Seen
Valid until: 12/30/2026
Sources
Related Entities
Related Claims
UnverifiedThe cost breakthroughs of domestic models such as DeepSeek are attributable to the MoE sparse architecture, FlashAttention acceleration technology, and efficient distributed training scheduling strategies.78% similarUnverifiedDeepSeek open-sourced its trained model weights, training data recipe, and technical report.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/963628API
curl https://kongchang.com/api/v1/knowledge/claims/963628MCP
get_claim(id=963628)