change example comand to 32768 context
Browse filesnot sure the max non-yarn context of this model, so made it smaller than
my original 64k example
README.md
CHANGED
|
@@ -427,7 +427,7 @@ numactl -N "$SOCKET" -m "$SOCKET" \
|
|
| 427 |
./build/bin/llama-server \
|
| 428 |
--model "$model"\
|
| 429 |
--alias ubergarm/Ling-1T-GGUF \
|
| 430 |
-
--ctx-size
|
| 431 |
-fa -fmoe -ger \
|
| 432 |
-ctk q8_0 -ctv q8_0 \
|
| 433 |
-ub 4096 -b 4096 \
|
|
|
|
| 427 |
./build/bin/llama-server \
|
| 428 |
--model "$model"\
|
| 429 |
--alias ubergarm/Ling-1T-GGUF \
|
| 430 |
+
--ctx-size 32768 \
|
| 431 |
-fa -fmoe -ger \
|
| 432 |
-ctk q8_0 -ctv q8_0 \
|
| 433 |
-ub 4096 -b 4096 \
|