emrgnt-cmplxty commited on
Commit
753ff5e
1 Parent(s): f288e10

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +17 -8
README.md CHANGED
@@ -5,7 +5,7 @@ size_categories:
5
  - 1B<n<10B
6
  task_categories:
7
  - text-generation
8
- pretty_name: OpenWebSearch-V1
9
  configs:
10
  - config_name: default
11
  data_files:
@@ -38,9 +38,9 @@ dataset_info:
38
 
39
  ### Getting Started
40
 
41
- The OpenWebSearch-V1 dataset includes full embeddings for over 50 million high-quality documents. This extensive collection encompasses the majority of content from sources like Arxiv, Wikipedia, Project Gutenberg, and includes quality-filtered CC data.
42
 
43
- To access and utilize the OpenWebSearch-1B dataset, you can stream it via HuggingFace with the following Python code:
44
 
45
  ```python
46
  from datasets import load_dataset
@@ -53,11 +53,11 @@ ds = load_dataset("SciPhi/OpenWebSearch-V1", data_files="arxiv/*", streaming=Tru
53
 
54
  ---
55
 
56
- A full set of scripts to recreate the dataset from scratch can be found [here](https://github.com/SciPhi/OpenWebSearch).
57
 
58
  ### Dataset Summary
59
 
60
- OpenWebSearch is divided into a number of categories, similar to RedPajama-V1.
61
 
62
 
63
  | Dataset | Token Count |
@@ -89,13 +89,22 @@ The raw dataset structure is as follows:
89
  }
90
  ```
91
 
92
- The indexed dataset is structured as a qdrant database dump, each entry has meta data {"url", "vector"}.
93
 
94
  ## Dataset Creation
95
 
96
- This dataset was created to allow make humanities most important knowledge locally searchable. It was created by filtering, cleaning, and augmenting locally publicly available datasets.
97
 
98
- The embedding vectors have been indexed and made searchable via a qdrant database.
 
 
 
 
 
 
 
 
 
99
 
100
  ### Source Data
101
 
 
5
  - 1B<n<10B
6
  task_categories:
7
  - text-generation
8
+ pretty_name: AgentSearch-V1
9
  configs:
10
  - config_name: default
11
  data_files:
 
38
 
39
  ### Getting Started
40
 
41
+ The AgentSearch-V1 dataset includes over one billion embeddings sourced from over 50 million high-quality documents. This extensive collection encompasses the majority of content from sources like Arxiv, Wikipedia, Project Gutenberg, and includes quality-filtered CC data.
42
 
43
+ To access and utilize the AgentSearch-V1 dataset, you can stream it via HuggingFace with the following Python code:
44
 
45
  ```python
46
  from datasets import load_dataset
 
53
 
54
  ---
55
 
56
+ A full set of scripts to recreate the dataset from scratch can be found [here](https://github.com/SciPhi/agent-search).
57
 
58
  ### Dataset Summary
59
 
60
+ We take a similar approach to RedPajama-v1 and divide AgentSearch into a number of categories.
61
 
62
 
63
  | Dataset | Token Count |
 
89
  }
90
  ```
91
 
92
+ The indexed dataset can be downloaded directly and is structured as a qdrant database dump, each entry has meta data {"url", "vector"}. In addition, there is a corresponding sqlite dataset which contains the mapping from urls onto embeddings, text chunks, and other metadata.
93
 
94
  ## Dataset Creation
95
 
96
+ This dataset was created as a step towards making humanities most important knowledge locally searchable and LLM optimal. It was created by filtering, cleaning, and augmenting locally publicly available datasets.
97
 
98
+ To cite our work, please use the following:
99
+
100
+ ```
101
+ @software{SciPhi2023AgentSearch,
102
+ author = {SciPhi},
103
+ title = {AgentSearch [ΨΦ]: A Comprehensive Agent-First Framework and Dataset for Webscale Search},
104
+ year = {2023},
105
+ url = {https://github.com/SciPhi-AI/agent-search}
106
+ }
107
+ ```
108
 
109
  ### Source Data
110