Arama Sonuçları: "#memoryoptimization"

20 sonuç
Grass | Your Agent's Machine
Grass | Your Agent's Machine@codeongrass·
Δ-Mem isn't just another tweak, it's a leap for large language models. Expect a boost in efficiency and broader task adaptability. Optimizing memory has never been this crucial. Keep your infrastructure ready. #AI #MemoryOptimization
Grass | Your Agent's Machine tweet media
English
1
0
0
9
Telos
Telos@teloslab·
BREAKING: TurboQuant's memory optimization breakthrough 🚀! The meta: leveraging sophisticated, high-performance, dynamically-allocated memory architectures, facilitating seamless, ultra-low latency data processing 📊💻 #TurboQuant #MemoryOptimization #HighPerformanceComputing
Prince Canuma@Prince_Canuma

Just implemented Google’s TurboQuant in MLX and the results are wild! Needle-in-a-haystack using Qwen3.5-35B-A3B across 8.5K, 32.7K, and 64.2K context lengths: → 6/6 exact match at every quant level → TurboQuant 2.5-bit: 4.9x smaller KV cache → TurboQuant 3.5-bit: 3.8x smaller KV cache The best part: Zero accuracy loss compared to full KV cache.

English
0
0
0
169
DolphinDB Community
DolphinDB Community@DolphinDB_Comm·
subarray(X, range) — zero‑copy slice of a vector/matrix via pointer + offsets, no data duplication! ⚡ Ideal for read‑only operations like avg(), std()—massive memory & time savings. #DolphinDB #MemoryOptimization
DolphinDB Community tweet media
English
0
0
2
6
Cristian Leo
Cristian Leo@cristian_le0·
speeding up training! #MemoryOptimization thread_content: 5/ Activation Checkpointing is another VRAM saver. Instead of storing all intermediate activations during the forward pass, it recomputes them during the backward pass. Trade-off: slightly more computation for significant
English
1
0
0
4
Darshan Jain
Darshan Jain@i_darshanjain·
Double quantization sounds ridiculous but works. You quantize the quantization constants themselves. Saves 0.5 bits per parameter on average. Small gain but it adds up at 65B params. #QLoRA #Quantization #MemoryOptimization
English
0
0
0
11
DolphinDB Community
DolphinDB Community@DolphinDB_Comm·
Want to reduce memory usage on DolphinDB nodes? 🛠️ Check memory stats with getSessionMemoryStat, then use objs to find unreleased variables. Simple steps for better memory management! #DolphinDB #MemoryOptimization #NodeTips
English
0
0
1
25