answerai-colbert-small-v1

answerai-colbert-small-v1 是 Answer.AI 推出的全新概念验证模型，展示了多向量模型在使用全新的 JaColBERTv2.5 训练方案 和一些额外优化后所能达到的强大性能，即使仅有 3300 万参数。

虽然它的尺寸与 MiniLM 相当，但在常见基准测试上超越了所有此前同尺寸的模型，甚至超越了更大的流行模型，如 e5-large-v2 或 bge-base-en-v1.5。

有关此模型或其训练方式的更多信息，请访问公告博客文章。

架构

<iframe src="https://brianronan-netron-docker.hf.space?repo_id=brianronan/answerai-colbert-small-v1&filename=model.safetensors" style="width: 95%; height: 70vh; border: none;"

</iframe>

使用方法

安装

该模型是为即将到来的 RAGatouille 重构而设计的。不过，它与所有最新的 ColBERT 实现兼容！

要使用它，您可以使用 Stanford ColBERT 库或 RAGatouille。您可以通过简单地运行以下命令来安装其中一个或两者。

pip install --upgrade ragatouille
pip install --upgrade colbert-ai

如果您有兴趣将此模型用作重排序器（它的性能远超同尺寸的交叉编码器！），可以通过 rerankers 库来实现：

pip install --upgrade rerankers[transformers]

Rerankers

from rerankers import Reranker

ranker = Reranker("answerdotai/answerai-colbert-small-v1", model_type='colbert')
docs = ['Hayao Miyazaki is a Japanese director, born on [...]', 'Walt Disney is an American author, director and [...]', ...]
query = 'Who directed spirited away?'
ranker.rank(query=query, docs=docs)

RAGatouille

from ragatouille import RAGPretrainedModel

RAG = RAGPretrainedModel.from_pretrained("answerdotai/answerai-colbert-small-v1")

docs = ['Hayao Miyazaki is a Japanese director, born on [...]', 'Walt Disney is an American author, director and [...]', ...]

RAG.index(documents, index_name="ghibli")

query = 'Who directed spirited away?'
results = RAG.search(query)

Stanford ColBERT

索引

from colbert import Indexer
from colbert.infra import Run, RunConfig, ColBERTConfig

INDEX_NAME = "DEFINE_HERE" 

if __name__ == "__main__":
    config = ColBERTConfig(
        doc_maxlen=512,
        nbits=2
    )
    indexer = Indexer(
        checkpoint="answerdotai/answerai-colbert-small-v1",
        config=config,
    )
    docs = ['Hayao Miyazaki is a Japanese director, born on [...]', 'Walt Disney is an American author, director and [...]', ...]

    indexer.index(name=INDEX_NAME, collection=docs)

查询

from colbert import Searcher
from colbert.infra import Run, RunConfig, ColBERTConfig

INDEX_NAME = "THE_INDEX_YOU_CREATED" 
k = 10 

if __name__ == "__main__":
    config = ColBERTConfig(
        query_maxlen=32 # Adjust as needed, we recommend the nearest higher multiple of 16 to your query
    )
    searcher = Searcher(
        index=index_name, 
        config=config
    ) 
    query = 'Who directed spirited away?'
    results = searcher.search(query, k=k)

提取向量

最后，如果您想提取单个向量，可以这样使用模型：

from colbert.modeling.checkpoint import Checkpoint

ckpt = Checkpoint("answerdotai/answerai-colbert-small-v1", colbert_config=ColBERTConfig())
embedded_query = ckpt.queryFromText(["Who dubs Howl's in English?"], bsize=16)

结果

与单向量模型对比

与 ColBERTv2.0 对比

引用

我们很可能会在将来发布技术报告。在此期间，如果您使用此模型或其他遵循 JaColBERTv2.5 方案的模型，并希望给予我们认可，请引用 JaColBERTv2.5 期刊预印本：

@article{clavie2024jacolbertv2,
  title={JaColBERTv2.5: Optimising Multi-Vector Retrievers to Create State-of-the-Art Japanese Retrievers with Constrained Resources},
  author={Clavi{\'e}, Benjamin},
  journal={arXiv preprint arXiv:2407.20750},
  year={2024}
}