Skip to content

A LLM trained only on data from certain time periods to reduce modern bias

Python

Revise README with language links and project details

Hayk Grigoriancbdfd927f3ed
diff --git a/README.md b/README.md
index 81d8221..4537714 100644
--- a/README.md
+++ b/README.md
@@ -77,6 +77,9 @@ At first I assumed that a protest might have coincidentally taken place the same
### **v2mini-eval1**
+Trained using a 15GB sample from v2's 90GB dataset.
+Model trained to 10K steps only.
+
There was an issue with tokenization that causes output to look like this:
PROMPT:
@@ -191,5 +194,6 @@ GPU: A100 SXM rented
+