|
|
|
|
|
|
|
datasets: |
|
- namespace: lilac |
|
name: imdb |
|
source: |
|
dataset_name: imdb |
|
source_name: huggingface |
|
embeddings: |
|
- path: text |
|
embedding: gte-small |
|
signals: |
|
- path: text |
|
signal: |
|
signal_name: near_dup |
|
- path: text |
|
signal: |
|
signal_name: pii |
|
- path: text |
|
signal: |
|
signal_name: lang_detection |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- text |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: open-asssistant-conversations |
|
source: |
|
dataset_name: OpenAssistant/oasst1 |
|
source_name: huggingface |
|
embeddings: |
|
- path: text |
|
embedding: gte-small |
|
signals: |
|
- path: text |
|
signal: |
|
signal_name: near_dup |
|
- path: text |
|
signal: |
|
signal_name: pii |
|
- path: text |
|
signal: |
|
signal_name: lang_detection |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- text |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: wikitext-2-raw-v1 |
|
source: |
|
dataset_name: wikitext |
|
config_name: wikitext-2-raw-v1 |
|
source_name: huggingface |
|
embeddings: |
|
- path: text |
|
embedding: gte-small |
|
signals: |
|
- path: text |
|
signal: |
|
signal_name: near_dup |
|
- path: text |
|
signal: |
|
signal_name: pii |
|
- path: text |
|
signal: |
|
signal_name: lang_detection |
|
- path: text |
|
signal: |
|
signal_name: text_statistics |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
settings: |
|
ui: |
|
media_paths: |
|
- text |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: squad_v2 |
|
source: |
|
dataset_name: squad_v2 |
|
source_name: huggingface |
|
embeddings: |
|
- path: context |
|
embedding: gte-small |
|
signals: |
|
- path: context |
|
signal: |
|
signal_name: near_dup |
|
- path: context |
|
signal: |
|
signal_name: pii |
|
- path: context |
|
signal: |
|
signal_name: lang_detection |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: context |
|
signal: |
|
signal_name: text_statistics |
|
- path: question |
|
signal: |
|
signal_name: near_dup |
|
- path: question |
|
signal: |
|
signal_name: pii |
|
- path: question |
|
signal: |
|
signal_name: lang_detection |
|
- path: question |
|
signal: |
|
signal_name: text_statistics |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
signal_name: near_dup |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
signal_name: pii |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
signal_name: lang_detection |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
signal_name: text_statistics |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: |
|
- answers |
|
- text |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
settings: |
|
ui: |
|
media_paths: |
|
- context |
|
- question |
|
- - answers |
|
- text |
|
- '*' |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: databricks-dolly-15k-curated-en |
|
source: |
|
dataset_name: argilla/databricks-dolly-15k-curated-en |
|
source_name: huggingface |
|
embeddings: |
|
- path: original-context |
|
embedding: gte-small |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
embedding: gte-small |
|
signals: |
|
- path: original-instruction |
|
signal: |
|
signal_name: near_dup |
|
- path: original-instruction |
|
signal: |
|
signal_name: pii |
|
- path: original-instruction |
|
signal: |
|
signal_name: lang_detection |
|
- path: original-instruction |
|
signal: |
|
signal_name: text_statistics |
|
- path: original-context |
|
signal: |
|
signal_name: near_dup |
|
- path: original-context |
|
signal: |
|
signal_name: pii |
|
- path: original-context |
|
signal: |
|
signal_name: lang_detection |
|
- path: original-context |
|
signal: |
|
signal_name: text_statistics |
|
- path: original-response |
|
signal: |
|
signal_name: near_dup |
|
- path: original-response |
|
signal: |
|
signal_name: pii |
|
- path: original-response |
|
signal: |
|
signal_name: lang_detection |
|
- path: original-response |
|
signal: |
|
signal_name: text_statistics |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: near_dup |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: pii |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: lang_detection |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: text_statistics |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: near_dup |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: pii |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: lang_detection |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: text_statistics |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: near_dup |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: pii |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: lang_detection |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
signal_name: text_statistics |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: original-instruction |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: original-context |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: original-response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: |
|
- new-instruction |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: |
|
- new-context |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: |
|
- new-response |
|
- value |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
settings: |
|
ui: |
|
media_paths: |
|
- original-instruction |
|
- original-context |
|
- original-response |
|
- - new-instruction |
|
- value |
|
- '*' |
|
- - new-context |
|
- value |
|
- '*' |
|
- - new-response |
|
- value |
|
- '*' |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: piqa |
|
source: |
|
dataset_name: piqa |
|
source_name: huggingface |
|
embeddings: |
|
- path: goal |
|
embedding: gte-small |
|
- path: sol1 |
|
embedding: gte-small |
|
- path: sol2 |
|
embedding: gte-small |
|
signals: |
|
- path: goal |
|
signal: |
|
signal_name: near_dup |
|
- path: goal |
|
signal: |
|
signal_name: pii |
|
- path: goal |
|
signal: |
|
signal_name: lang_detection |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: goal |
|
signal: |
|
signal_name: text_statistics |
|
- path: sol1 |
|
signal: |
|
signal_name: near_dup |
|
- path: sol1 |
|
signal: |
|
signal_name: pii |
|
- path: sol1 |
|
signal: |
|
signal_name: lang_detection |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: sol1 |
|
signal: |
|
signal_name: text_statistics |
|
- path: sol2 |
|
signal: |
|
signal_name: near_dup |
|
- path: sol2 |
|
signal: |
|
signal_name: pii |
|
- path: sol2 |
|
signal: |
|
signal_name: lang_detection |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: sol2 |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- sol1 |
|
- sol2 |
|
- goal |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: OpenOrca-100k |
|
source: |
|
dataset_name: Open-Orca/OpenOrca |
|
sample_size: 100000 |
|
source_name: huggingface |
|
embeddings: |
|
- path: question |
|
embedding: gte-small |
|
- path: response |
|
embedding: gte-small |
|
signals: |
|
- path: question |
|
signal: |
|
signal_name: near_dup |
|
- path: question |
|
signal: |
|
signal_name: pii |
|
- path: question |
|
signal: |
|
signal_name: lang_detection |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
signal_name: text_statistics |
|
- path: response |
|
signal: |
|
signal_name: near_dup |
|
- path: response |
|
signal: |
|
signal_name: pii |
|
- path: response |
|
signal: |
|
signal_name: lang_detection |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: response |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- question |
|
- response |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: opus100-en-es-validation |
|
source: |
|
dataset_name: opus100 |
|
config_name: en-es |
|
split: validation |
|
source_name: huggingface |
|
embeddings: |
|
- path: |
|
- translation |
|
- en |
|
embedding: gte-small |
|
- path: |
|
- translation |
|
- es |
|
embedding: gte-small |
|
signals: |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
signal_name: near_dup |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
signal_name: pii |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
signal_name: lang_detection |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- en |
|
signal: |
|
signal_name: text_statistics |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
signal_name: near_dup |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
signal_name: pii |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
signal_name: lang_detection |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: |
|
- translation |
|
- es |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- - translation |
|
- es |
|
- - translation |
|
- en |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: mmlu_professional_law |
|
source: |
|
dataset_name: cais/mmlu |
|
config_name: professional_law |
|
source_name: huggingface |
|
embeddings: |
|
- path: question |
|
embedding: gte-small |
|
- path: |
|
- choices |
|
- '*' |
|
embedding: gte-small |
|
signals: |
|
- path: question |
|
signal: |
|
signal_name: near_dup |
|
- path: question |
|
signal: |
|
signal_name: pii |
|
- path: question |
|
signal: |
|
signal_name: lang_detection |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: question |
|
signal: |
|
signal_name: text_statistics |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
signal_name: near_dup |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
signal_name: pii |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
signal_name: lang_detection |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: |
|
- choices |
|
- '*' |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- question |
|
- - choices |
|
- '*' |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: pile-of-law-r-legaladvice |
|
source: |
|
dataset_name: pile-of-law/pile-of-law |
|
config_name: r_legaladvice |
|
source_name: huggingface |
|
embeddings: |
|
- path: text |
|
embedding: gte-small |
|
signals: |
|
- path: text |
|
signal: |
|
signal_name: near_dup |
|
- path: text |
|
signal: |
|
signal_name: pii |
|
- path: text |
|
signal: |
|
signal_name: lang_detection |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- text |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: science-qa-derek-thomas |
|
source: |
|
dataset_name: derek-thomas/ScienceQA |
|
source_name: huggingface |
|
embeddings: |
|
- path: lecture |
|
embedding: gte-small |
|
signals: |
|
- path: lecture |
|
signal: |
|
signal_name: near_dup |
|
- path: lecture |
|
signal: |
|
signal_name: pii |
|
- path: lecture |
|
signal: |
|
signal_name: lang_detection |
|
- path: lecture |
|
signal: |
|
signal_name: text_statistics |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: lecture |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
settings: |
|
ui: |
|
media_paths: |
|
- lecture |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: enron-emails |
|
source: |
|
dataset_name: EleutherAI/pile |
|
config_name: enron_emails |
|
sample_size: 100000 |
|
source_name: huggingface |
|
embeddings: |
|
- path: text |
|
embedding: gte-small |
|
signals: |
|
- path: text |
|
signal: |
|
signal_name: near_dup |
|
- path: text |
|
signal: |
|
signal_name: pii |
|
- path: text |
|
signal: |
|
signal_name: lang_detection |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: text |
|
signal: |
|
signal_name: text_statistics |
|
settings: |
|
ui: |
|
media_paths: |
|
- text |
|
preferred_embedding: gte-small |
|
- namespace: lilac |
|
name: the_movies_dataset |
|
source: |
|
filepaths: |
|
- https://storage.googleapis.com/lilac-data/datasets/the_movies_dataset/the_movies_dataset.csv |
|
source_name: csv |
|
embeddings: |
|
- path: overview |
|
embedding: gte-small |
|
signals: |
|
- path: overview |
|
signal: |
|
signal_name: near_dup |
|
- path: overview |
|
signal: |
|
signal_name: pii |
|
- path: overview |
|
signal: |
|
signal_name: lang_detection |
|
- path: overview |
|
signal: |
|
signal_name: text_statistics |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: legal-termination |
|
signal_name: concept_score |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: negative-sentiment |
|
signal_name: concept_score |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: non-english |
|
signal_name: concept_score |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: positive-sentiment |
|
signal_name: concept_score |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: profanity |
|
signal_name: concept_score |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: question |
|
signal_name: concept_score |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: source-code |
|
signal_name: concept_score |
|
- path: overview |
|
signal: |
|
embedding: gte-small |
|
namespace: lilac |
|
concept_name: toxicity |
|
signal_name: concept_score |
|
settings: |
|
ui: |
|
media_paths: |
|
- overview |
|
preferred_embedding: gte-small |
|
|
