mradermacher
/

model_requests

Model card Files Files and versions Community

mradermacher commited on Apr 26

Commit

510a000

•

1 Parent(s): c345c2c

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -106,7 +106,7 @@ uploads are small.
 Through a combination of these ingenuous tricks:
-1. I am not above using a low quant (e.g. Q4_K_S, IQ3_XXS or even Q2_K), reducing the size of the model.
 2. An nvme drive is "only" 25-50 times slower than RAM. I lock the first 80GB of the model in RAM, and
    then stream the remaining data from disk for every iteration.
 3. Patience.

 Through a combination of these ingenuous tricks:
+1. I am not above using a low quant (e.g. Q4_K_S, IQ3_XS or even Q2_K), reducing the size of the model.
 2. An nvme drive is "only" 25-50 times slower than RAM. I lock the first 80GB of the model in RAM, and
    then stream the remaining data from disk for every iteration.
 3. Patience.