diff options
| -rwxr-xr-x | convert-falcon-hf-to-gguf.py | 1 | ||||
| -rwxr-xr-x | scripts/run-all-ppl.sh | 2 |
2 files changed, 2 insertions, 1 deletions
diff --git a/convert-falcon-hf-to-gguf.py b/convert-falcon-hf-to-gguf.py index 43e20849..411cbf68 100755 --- a/convert-falcon-hf-to-gguf.py +++ b/convert-falcon-hf-to-gguf.py @@ -107,6 +107,7 @@ if "n_head_kv" in hparams: else: gguf_writer.add_head_count_kv(1) gguf_writer.add_layer_norm_eps(hparams["layer_norm_epsilon"]) +gguf_writer.add_file_type(ftype) # TOKENIZATION diff --git a/scripts/run-all-ppl.sh b/scripts/run-all-ppl.sh index bdbbb19c..366d0866 100755 --- a/scripts/run-all-ppl.sh +++ b/scripts/run-all-ppl.sh @@ -1,7 +1,7 @@ #!/bin/bash qnt=(f16 q8_0 q6_k q5_k q5_1 q5_0 q4_k q4_1 q4_0 q3_k q2_k) -args="--no-mmap -ngl 999 -t 8" +args="-ngl 999 -t 8" if [ -z "$1" ]; then echo "usage: $0 <model> [qnt] [args]" |