Aeonthic/OpenLumen-2-Pulsar
Full fine-tune (fewer layers) vs google/gemma-4-26B-A4B. Aeonthic/OpenLumen-2-Pulsar matches part of google/gemma-4-26B-A4B (657 of its 1013 tensors; the rest of google/gemma-4-26B-A4B has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 30 of 30 layers.
Tensors changed: Embeddings 1/1, MLP 150/150, Norms 211/271, other 88/120, Attention 115/115
2 derivatives checked
- mradermacher/OpenLumen-2-Pulsar-i1-GGUF: Quantization
- mradermacher/OpenLumen-2-Pulsar-GGUF: Quantization