Anthropic published a paper applying the cognitive-science concept of a global workspace to Claude, revealing attention as working memory. Meanwhile, robotics releases from Hugging Face, Photoroom, an…
Photoroom published the fourth part of its PRX series, detailing the data strategy behind its image generation model. The team assembled a diverse pre-training dataset from public and internal sources…
Photoroom engineers found that classifier-free guidance scales above 7.5 caused oversaturation and clipped highlights in their SDXL product renders. They fixed the issue by adding CFG rescale at 0.7 a…
Photoroom reduced inference latency of its convolutional segmentation backbone by 22% on A100 GPUs by switching to PyTorch's channels-last memory format. The change required only four lines of code an…
Photoroom reduced its vision-language model inference costs by approximately 62% within three weeks by deploying Bifrost as a semantic caching layer in front of the VLM step of its product-photo diffu…
Photoroom's SDXL LoRA fine-tuning for a product photography model silently corrupted its adapter weights over six days due to a bf16 gradient accumulation issue. The custom training loop, forked from …
At Google I/O ‘26, Google announced new cost-effective frontier models for startups, including Gemini 3.5 Flash for agentic tasks and Gemini Omni for dynamic video content creation. The company also i…