📰00Read later0Read MoreLoRA on a 16 GB card: what fits and what does notDEV Community: pytorch·Ivan Zdravkov·19 days ago#tRB8q6Kx#devtopytorch#devto#dev#memory#card+5 more🧰Tag tools✨Add tagA 7B model is the honest ceiling for LoRA fine-tuning on a 16 GB consumer card, not because anything...15s
📰00Read later0Read MoreDurable execution without history replayHacker News·2 months ago#4m7urkSN#trigora#checkpointing#tcc#concurrency#article+1 more🧰Tag tools✨Add tagMost durable execution systems recover by replaying retained history. TCC checkpoints the program continuation instead, so recovery need not reconstruct the present from the past.15s
🖼️00Read later0Read MoreInternals: How PyTorch 2.5 and TensorFlow 2.17 Implement Gradient Checkpointing for LLM Fine-TuningDEV Community·ANKUSH CHOUDHARY JOHAL·6 months ago#vUJe0eHH#pytorch#tensorflow#internals#implement#checkpointing+5 more🧰Tag tools✨Add tagFrom Dev.to - tensorflow: Internals: How PyTorch 2.5 and TensorFlow 2.17 Implement Gradient Checkpointing for LLM Fine-Tuning15s