Unverified50% confidenceFactExact time
SlothOps uses Thinking Machines' Tinker SDK to perform LoRA fine-tuning on meta-llama/Llama-3-8B-Instruct with rank 16 and alpha 32, applied to the four projection layers q/v/k/o.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
Related Entities
Related Claims
Unverified许多最新的VLM（如PaLI-X、LLaVA-1.6等）已将视觉编码器从CLIP切换为SigLIP62% similarUnverifiedThe project uses Tinker to perform supervised fine-tuning of Qwen3-8B.62% similarUnverified第一代Qwen起步上下文窗口为2048，通过NTK插值外推到8192，训练使用Flash Attention、AdamW优化器和BF16混合精度61% similarUnverifiedMistral Large 3 supports a native context window of 512,000 tokens, combining block sparse attention with hardware-calibrated rotary position embeddings (RoPE).61% similarUnverifiedThe model architecture is a cross-attention DiT (896 width × 16 layers) using 2D RoPE, QK-norm, SwiGLU, and adaLN-single.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1072940API
curl https://kongchang.com/api/v1/knowledge/claims/1072940MCP
get_claim(id=1072940)