From 9cd27b2f91111e7ff991cfd464bccc3dc9ffa86a Mon Sep 17 00:00:00 2001
From: NanoCode012 <kevinvong@rocketmail.com>
Date: Tue, 16 Jan 2024 09:47:33 +0900
Subject: [PATCH] fix(readme): clarify custom user prompt [no-ci] (#1124)

* fix(readme): clarify custom user prompt

* chore: update example to show use case of setting field
---
 README.md | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 2bd3d52814..94a0c8f8b4 100644
--- a/README.md
+++ b/README.md
@@ -374,7 +374,7 @@ Have dataset(s) in one of the following format (JSONL recommended):
 For a dataset that is preprocessed for instruction purposes:
 
 ```json
-{"instruction": "...", "output": "..."}
+{"input": "...", "output": "..."}
 ```
 
 You can use this example in your YAML config:
@@ -385,6 +385,8 @@ datasets:
     type:
       system_prompt: ""
       field_system: system
+      field_instruction: input
+      field_output: output
       format: "[INST] {instruction} [/INST]"
       no_input_format: "[INST] {instruction} [/INST]"
 ```
@@ -577,10 +579,10 @@ datasets:
     field_human: # Optional[str]. Human key to use for conversation.
     field_model: # Optional[str]. Assistant key to use for conversation.
 
-  # Custom user prompt
+  # Custom user instruction prompt
   - path: repo
     type:
-      # The below are defaults. only set what's needed.
+      # The below are defaults. only set what's needed if you use a different column name.
       system_prompt: ""
       system_format: "{system}"
       field_system: system
@@ -589,6 +591,7 @@ datasets:
       field_output: output
 
       # Customizable to be single line or multi-line
+      # Use {instruction}/{input} as key to be replaced
       # 'format' can include {input}
       format: |-
         User: {instruction} {input}