Datasets:

ibm-granite
/

GneissWeb

Languages:

ArXiv:

License:

Dataset card Files Files and versions

bhatta1 commited on Feb 13

Commit

f3f4470

·

verified ·

1 Parent(s): 3ed65d8

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -198,7 +198,7 @@ Guided by the distributions of TokensPerChar and TokensPerByte in different cate
 In Figure 18, we show the progression of  accuracy with training for High Signal Tasks for 1.4 billion parameter model for 35 billion tokens.  We see that for both datasets compared, the accuracy increases over time and the accuracy of the dataset with Extreme_tokenized quality filter at 52.78  is higher than the baseline at 51.94.
 **Figure 18:** Ablation experiment comparing Extreme_tokenized Filter against the FineWeb.V1.1 baseline at 1.4 Billion model size for 35 Billion tokens

 In Figure 18, we show the progression of  accuracy with training for High Signal Tasks for 1.4 billion parameter model for 35 billion tokens.  We see that for both datasets compared, the accuracy increases over time and the accuracy of the dataset with Extreme_tokenized quality filter at 52.78  is higher than the baseline at 51.94.
+<img src="Extreme.png" alt="Extreme.png" style="width:1000px;"/>
 **Figure 18:** Ablation experiment comparing Extreme_tokenized Filter against the FineWeb.V1.1 baseline at 1.4 Billion model size for 35 Billion tokens