These are some of the retrieval training datasets used for training RaDeR models, sonsisting of different types of query combinations.
RaDeR
non-profit
AI & ML interests
None defined yet.
Recent Activity
View all activity
Organization Card
Paper: RaDeR: Reasoning-aware Dense Retrieval Models
RaDeR are a series of reasoning based dense retrievers and rerankers, trained using data derived from mathematical problem solving using large language models.
Github Link: https://github.com/Debrup-61/RaDeR.
Work from Center of Intelligent Information Retrieval (CIIR), UMass Amherst.
models
8
Raderspace/RaDeR_Qwen25_3B_NuminaMath_MATH_allquerytypes
Feature Extraction
•
3B
•
Updated
•
2
Raderspace/RaDeR_Qwen25-7B_NuminaMath_MATH_allquerytypes
Feature Extraction
•
7B
•
Updated
•
6
Raderspace/RaDeR_Qwen_25_7B_instruct_MATH_LLMq_CoT_lexical
Feature Extraction
•
7B
•
Updated
•
20
Raderspace/RaDeR_gte_Qwen2-7B_MATH_LLMq_CoT_lexical
Feature Extraction
•
7B
•
Updated
•
3
Raderspace/RaDeR_Qwen25-14B_NuminaMath_MATH_allquerytypes
Feature Extraction
•
14B
•
Updated
•
1
Raderspace/reranker_Qwen25_7B_NuminaMath_MATH_allquerytypes
Updated
Raderspace/lora_RaDeR_gte_Qwen2-7B_MATH_LLMq_CoT_lexical
Updated
Raderspace/lora_RaDeR_Qwen25-7B_instruct_MATH_LLMq_CoT_lexical
Updated