Bobbie-model Patched Jun 2026

| Benchmark | Bobbie-7B | Llama-3-8B | Mistral-7B | |-----------|-----------|------------|------------| | MMLU (5-shot) | 64.2 | 66.7 | 63.9 | | GSM8K (8-shot) | 52.8 | 54.9 | 50.3 | | HumanEval (pass@1) | 32.5 | 34.2 | 31.8 | | | 82.3 | 67.1 | 71.4 | | Inference tokens/sec | 98 | 72 | 88 |

Bobbie doesn't require a separate "long context" fine-tune. You can feed it 30k tokens out of the box. bobbie-model

This model explores how various scheduling practices affect patient wait times, especially in environments with many walk-in patients. | Benchmark | Bobbie-7B | Llama-3-8B | Mistral-7B

In this post, we’ll strip down the architecture, analyze its training data strategy, and run benchmarks against comparable 7B models. In this post, we’ll strip down the architecture,

Why? To keep the KV cache small during long document processing. The model first "scouts" the context window, then commits to full attention.