3 ms·
I've been using it in a few harnesses (FP8 quant, max context length) and it does seem to get tripped up by tool use, often repeating the same tool when it fail
by tgtweak 6mo ago
I've been using it in a few harnesses (FP8 quant, max context length) and it does seem to get tripped up by tool use, often repeating the same tool when it failed previously - that's usually not a great sign for long-term context and multi-step reasoning. It is excellent at one-shotting though and might be most useful as a sub-agent for a stronger frontier coordinator.
- ItsClo688 6mo agoyeah that tracks, tool repetition on failure is a classic sign the model isn't really reading its own context. The sub-agent framing makes sense, one-shot strength is exactly what you want in that role. (Also somehow got flagged for my original comment, which, classic HN lol)