NVIDIA Releases Nemotron-Labs-TwoTower: an Open-Weight Diffusion Language Model Built on a Frozen Autoregressive Nemotron-3-Nano-30B-A3B Backbone

NVIDIA has released Nemotron-Labs-TwoTower, a diffusion language model built on a pretrained autoregressive backbone. It ships as open weights under the NVIDIA Nemotron Open Model License. The release targets a throughput bottleneck in text generation. Autoregressive (AR) models decode one token at a time. That serial process caps generation throughput. Discrete diffusion language models take…

Read More

CUP (Common Useful Python): Building Reliable Python Workflows with Baidu’s Utility Toolkit

banner(“4. IN-MEMORY KV CACHE (cup.cache)”) try: from cup import cache kv = cache.KVCache(name=”demo”) kv.set({“user:1”: “alice”, “user:2”: “bob”}, expire_sec=2) kv.set({“config:flag”: “on”}, expire_sec=None) print(“size after sets :”, kv.size()) print(“get user:1 :”, kv.get(“user:1”)) print(“get missing key :”, kv.get(“nope”)) print(“sleeping 2.2s to let the 2s-TTL keys expire …”) time.sleep(2.2) print(“get user:1 (expired) :”, kv.get(“user:1”)) print(“get config:flag (eternal):”, kv.get(“config:flag”)) reclaimed…

Read More