andreaskoepf commited on
Commit
c6c1242
1 Parent(s): 8ba5a5a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +9 -11
README.md CHANGED
@@ -140,11 +140,10 @@ on the [open-assistant.io](https://www.open-assistant.io/) website until April,
140
  Conversations are exported as conversation trees with messages as nodes.
141
  The root node of a conversation tree is called the initial prompt. Each message can have
142
  multiple replies. Nodes with more than one reply can have a `rank` field indicating the
143
- user preference (the most preferred message has rank 0).
144
 
145
- All messages have a role which can either be "assistant" or "prompter". The roles in
146
- conversation threads from prompt to leaf node in a conversation tree are stricly alternating
147
- between "prompter" and "assistant".
148
 
149
  ### JSON Example: Message
150
 
@@ -183,7 +182,7 @@ Objects are stored without indentation on a single lines in the actual jsonl fil
183
 
184
  ### JSON Example: Conversation Tree
185
 
186
- For readability only a subset of properties are shown here.
187
 
188
  ```
189
  {
@@ -233,18 +232,17 @@ For readability only a subset of properties are shown here.
233
  }
234
  ```
235
 
236
-
237
  Please refer to [oasst-data](https://github.com/LAION-AI/Open-Assistant/tree/main/oasst-data) for
238
- details about the data structure and python code to read and write jsonl files containing oasst objects.
239
 
240
 
241
  ## Main Dataset Files
242
 
243
- Data is provided either as nested messages in conversation trees (extension `.trees.jsonl.gz`)
244
- or as flat list of messages (extension `.messages.jsonl.gz`).
245
 
246
- Full conversation trees can be reconstructed from flat messages using the `parent_id`
247
- and `message_id` properties to identify their parent-child relationship. The `message_tree_id`
248
  and `tree_state` properties (only present in flat messages files) can be used to find all
249
  all messages of a message tree or to select trees by their state.
250
 
 
140
  Conversations are exported as conversation trees with messages as nodes.
141
  The root node of a conversation tree is called the initial prompt. Each message can have
142
  multiple replies. Nodes with more than one reply can have a `rank` field indicating the
143
+ aggregated user preference (the most preferred message has rank 0).
144
 
145
+ All messages have a role property which can either be "assistant" or "prompter". The roles in
146
+ conversation threads from prompt to leaf node are stricly alternating between "prompter" and "assistant".
 
147
 
148
  ### JSON Example: Message
149
 
 
182
 
183
  ### JSON Example: Conversation Tree
184
 
185
+ For readability only a subset of the message properties is shown here.
186
 
187
  ```
188
  {
 
232
  }
233
  ```
234
 
 
235
  Please refer to [oasst-data](https://github.com/LAION-AI/Open-Assistant/tree/main/oasst-data) for
236
+ details about the data structure and Python code to read and write jsonl files containing oasst data objects.
237
 
238
 
239
  ## Main Dataset Files
240
 
241
+ Conversation data is provided either as nested messages in trees (extension `.trees.jsonl.gz`)
242
+ or as flat list (table) of messages (extension `.messages.jsonl.gz`).
243
 
244
+ Full conversation trees can be reconstructed from flat messages by using the `parent_id`
245
+ and `message_id` properties to identify the parent-child relationship of messages. The `message_tree_id`
246
  and `tree_state` properties (only present in flat messages files) can be used to find all
247
  all messages of a message tree or to select trees by their state.
248