Spaces:

mikeee
/

ggml-org-llama-cpp

Runtime error

mikeee commited on Jul 1

Commit

ff09653

verified ·

1 Parent(s): b8c6e4d

Update Dockerfile

try "-t", "4", \
"-tb", "4", \

Files changed (1) hide show

Dockerfile CHANGED Viewed

@@ -30,6 +30,8 @@ CMD [ \
   "--mmproj", "/gemma-4-E2B-it-mmproj.gguf", \
   "--port", "7860", \
   "--host", "0.0.0.0", \
   "-fa", "on", \
   "-ub", "128", \
   "-b", "512", \
@@ -37,7 +39,9 @@ CMD [ \
   "--spec-draft-n-max", "3" \
 ]
-# let llama-server autodetect -t -tb
 #  "-t", "2", \
 #  "-tb", "2", \

   "--mmproj", "/gemma-4-E2B-it-mmproj.gguf", \
   "--port", "7860", \
   "--host", "0.0.0.0", \
+  "-t", "4", \
+  "-tb", "4", \
   "-fa", "on", \
   "-ub", "128", \
   "-b", "512", \
   "--spec-draft-n-max", "3" \
 ]
+# let llama-server autodetect -t -tb: very slow
+#
+# 10 tps
 #  "-t", "2", \
 #  "-tb", "2", \