4 ms·
Umm, are these guys sending requests at a public inference endpoint or are they doing more sophisticated work like training etc?
by aitchnyu 6mo ago
Umm, are these guys sending requests at a public inference endpoint or are they doing more sophisticated work like training etc?
- pants2 6mo agoLock in is pretty easy these days. Just a dummy example, Claude models are trained on their `str_replace_based_edit_tool` edit tool[1] which is very different from OpenAI's `apply_patch` tool[2]. 1. https://platform.claude.com/docs/en/agents-and-tools/tool-use/text-editor-tool https://platform.claude.com/docs/en/agents-and-tools/tool-us... 2. https://developers.openai.com/api/docs/guides/tools-apply-patch https://developers.openai.com/api/docs/guides/tools-apply-pa...