We lifted 7 functions out of this paper's own repositories and ran 0 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.
| Repository | Role | Ran |
|---|---|---|
| gucci-j/chat-cve | canonical | 0 of 7 |
| Function | Status | Where it lives |
|---|---|---|
| add_transition | Not yet run | gucci-j/chat-cve/merging/src/methods/add_transition.py code served (permissive licence) · get_code("7a9ade35bc5412f1") |
| apply_swapping | Not yet run | gucci-j/chat-cve/merging/src/methods/apply_swapping.py code served (permissive licence) · get_code("6d1ce1df079d0080") |
| copy_emb | Not yet run | gucci-j/chat-cve/merging/src/methods/copy_emb.py code served (permissive licence) · get_code("b0ae80345791e4ad") |
| instantiate_model_by_mean | Not yet run | gucci-j/chat-cve/instantiation/src/methods/mean.py code served (permissive licence) · get_code("48cc29d2c1c88814") |
| lang_code_to_instruction | Not yet run | gucci-j/chat-cve/evaluation/src/gmmlu.py code served (permissive licence) · get_code("7909323ec4e084e4") |
| process_judge_response | Not yet run | gucci-j/chat-cve/evaluation/src/mtbench.py code served (permissive licence) · get_code("de8aed064593dbc8") |
| round_to_nearest_multiple | Not yet run | gucci-j/chat-cve/instantiation/src/methods/mean.py code served (permissive licence) · get_code("cee6083e09139c9a") |
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Vocabulary expansion (VE) is the de-facto approach to language adaptation of large language models (LLMs) by adding new tokens and continuing pre-training on target data. While this is effective for base models trained on unlabeled data, it poses challenges for chat models trained to follow instructions through labeled conversation data. Directly adapting the latter with VE on target unlabeled data may result in forgetting chat abilities. While ideal, target chat data is often unavailable or costly to create for low-resource languages, and machine-translated alternatives are not always effective. To address this issue, previous work proposed using a base and chat model from the same family. This method first adapts the base LLM with VE on target unlabeled data and then converts it to a chat model by adding a chat vector (CV) derived from the weight difference between the source base and chat models. We propose ElChat, a new language adaptation method for chat LLMs that adapts a chat model directly on target unlabeled data, without a base model. It elicits chat abilities by injecting information from the source chat model. ElChat offers more robust and competitive target language and safety performance while achieving superior English, chat, and instruction-following abilities compared to CV.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2412.11704")
get_code_for_paper("2412.11704")
have("2412.11704")
Connect an agent — have() is free.