Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
I have found how to train FLUX.1 lora，It seems like need 48G VRAM at least，here is a guidelines for training. https://t.co/8sUpYY1eDM @bdsqlsz @araminta_k @AIWarper @ZHOZHO672070 @camenduru @cubiq
Doing a full transformer finetune of FLUX.1 schnell at bf16 with T5 at 8bit is using ~ 75GB VRAM. https://t.co/8YzFKH9AwP https://t.co/zZUtMUBKzj
Testing training a LoRA for FLUX.1 ... 38.5gb VRAM!! Gradient checkpointing Rank 256 (Big, I know) T5 in 8bit Full BF16 It is a big boy. https://t.co/EMmiW6W7jC
Hyperbolic Labs has become the first to support the Llama 3.1 405B BASE model on OpenRouterAI, a significant development for open AI research. This model has been highly requested by AI researchers. Additionally, Hyperbolic Labs is working on integrating the Llama 3.1 405B Instruct bf16 model and has extended the context lengths of Llama 3.1 405B Base (fp8). The FLUX.1 image generation model is also being served, with users reporting improved performance and reduced memory usage on NVIDIA RTX 4090. Training and fine-tuning FLUX.1 requires substantial VRAM, with some configurations needing up to 75GB, 38.5GB, or 48G.