GLM-5.2 Weights Reduced by 423GB
EAccelerate_42 · x · 2026-07-13
The author claims to have reduced the size of GLM-5.2 by 423GB without altering the model:
- Reduced from 1403GB to 980GB
- Model size is 753B weights
- Results remain bit-for-bit exact
- No quantization and no retraining
The key insight is that the weights remain in a compressed state within VRAM, rather than reconstructing the full model first. The author mentions the full write-up and repo will be released in the next post.
More from Infra
- 34 open models self-hosted on a Blackwell cluster for open health evals — keunwoochoi · 2026-10-27
- Anthropic locked into up to $120.5B SpaceX compute deal — $3.1B per month from Dec 2026 — ns123abc · 2026-10-27
- Open-source Latch compiles PyTorch into training megakernels, 2.75x faster on H100 — sguada · 2026-10-27
- NVIDIA Claims HGX Rubin NVL8 Delivers 10x More Tokens per Megawatt Than B200 — thursdai_pod · 2026-10-27
- Commoditization dilemma: either human labor or GPUs become the commodity — _AustinCalvert_ · 2026-10-27
- DeepSpeed shows AutoEP/AutoTP/AutoSP parallelism gains, hosts office hours — PyTorch · 2026-10-27