metadata
language:
- en
pretty_name: ScienceWikiSmallChunk
tags:
- RAG
- Retrieval Augmented Generation
- Small Chunks
- Wikipedia
- Science
- Scientific
- Scientific Wikipedia
- Science Wikipedia
- 512 tokens
license: cc-by-sa-3.0
task_categories:
- text-generation
- text-classification
- question-answering
ScienceWikiSmallChunk
Processed version of millawell/wikipedia_field_of_science, prepared to be used in small context length RAG systems. Chunk length is tokenizer dependent, but each chunk should be around 512 tokens. Longer wikipedia pages have been split into smaller entries, with title added as a prefix.
There is also 256 tokens dataset available: Laz4rz/wikipedia_science_chunked_small_rag_256
If you wish to prepare some other chunk length:
- use millawell/wikipedia_field_of_science
- adapt chunker function:
def chunker_clean(results, example, length=512, approx_token=3, prefix=""):
if len(results) == 0:
regex_pattern = r'[\n\s]*\n[\n\s]*'
example = re.sub(regex_pattern, " ", example).strip().replace(prefix, "")
chunk_length = length * approx_token
if len(example) > chunk_length:
first = example[:chunk_length]
chunk = ".".join(first.split(".")[:-1])
if len(chunk) == 0:
chunk = first
rest = example[len(chunk)+1:]
results.append(prefix+chunk.strip())
if len(rest) > chunk_length:
chunker_clean(results, rest.strip(), length=length, approx_token=approx_token, prefix=prefix)
else:
results.append(prefix+rest.strip())
else:
results.append(prefix+example.strip())
return results