Update README.md
Browse files
README.md
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
**What is it?**
|
|
|
|
| 2 |
Data quantity and quality play a vital role in determining the performance of Large Language Models (LLMs). High-quality data, in particular, can significantly boost the LLM’s ability to generalize on a wide range of downstream tasks. To cater to the requirements of the Granite models, we focused on a goal to produce a 10T dataset, named, GneissWeb, that is of higher quality than all other datasets of similar size available. Gneiss, pronounced as nice, is a strong and durable rock that is used for building and construction.
|
| 3 |
|
| 4 |
The GneissWeb recipe consists of sharded exact substring deduplication and a judiciously constructed ensemble of quality filters. We present the key evaluations that guided our design choices and provide filtering thresholds that can be used to filter the dataset to match the token and quality needs of Stage-1 (early pre-training) or Stage-2 (annealing) datasets.
|
|
|
|
| 1 |
**What is it?**
|
| 2 |
+
|
| 3 |
Data quantity and quality play a vital role in determining the performance of Large Language Models (LLMs). High-quality data, in particular, can significantly boost the LLM’s ability to generalize on a wide range of downstream tasks. To cater to the requirements of the Granite models, we focused on a goal to produce a 10T dataset, named, GneissWeb, that is of higher quality than all other datasets of similar size available. Gneiss, pronounced as nice, is a strong and durable rock that is used for building and construction.
|
| 4 |
|
| 5 |
The GneissWeb recipe consists of sharded exact substring deduplication and a judiciously constructed ensemble of quality filters. We present the key evaluations that guided our design choices and provide filtering thresholds that can be used to filter the dataset to match the token and quality needs of Stage-1 (early pre-training) or Stage-2 (annealing) datasets.
|