Skip to content

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

Python

Update README.md

Hayk Grigorian86e5e02f61e2
diff --git a/README.md b/README.md
index 60826be..8a2ddc1 100644
--- a/README.md
+++ b/README.md
@@ -169,7 +169,8 @@ Dataset sizes:
- v0: ~187MB
- v0.5: ~435MB
- v1: ~6.25GB
-- v2mini-eval1: 15GB
+- v2mini-eval1: 15GB(sample from v2's 90GB)
+- v2: 90GB
## How large are the models ?
@@ -181,6 +182,10 @@ v1: 700M Parameters
v2mini-eval1: 300M Parameters
+v2mini-eval2: 200M Parameters
+
+v2: 1.2B Parameters
+
# Training Specs ?
# v0/v0.5
@@ -191,10 +196,13 @@ Ram: 16GB DDR5.
# v1
GPU: A100 SXM rented
-# v2mini-eval1
-
+# v2mini-eval1/eval2
GPU: A100 SXM rented
+# v2
+GPU: H100 SXM rented
+
+