From 8802d63c935a708bf01e9f7246b0e7dc1242ebc7 Mon Sep 17 00:00:00 2001
From: ltoniazzi <lt6ga@protonmail.com>
Date: Mon, 10 Jun 2024 22:48:39 +0100
Subject: [PATCH] first notes

---
 NOTES.md              | 7 +++++++
 convert-hf-to-gguf.py | 2 +-
 2 files changed, 8 insertions(+), 1 deletion(-)
 create mode 100644 NOTES.md

diff --git a/NOTES.md b/NOTES.md
new file mode 100644
index 000000000..9deb5bcd7
--- /dev/null
+++ b/NOTES.md
@@ -0,0 +1,7 @@
+# TODOs
+
+1. How to debug mat_mul (run tests in cpp?)
+2. How to wrap the suggestion from lauren on matmul (need to see how to find the llora info to pick up). Something about lora being loaded in the context? How to pick a specifi LoRA
+3. check the PR "It was removed in [#7204](https://github.com/ggerganov/llama.cpp/pull/7204). `convert-lora-to-ggml.py` seems to write  loras to gguf witouth the model? Should check the train script and see how they match lora with base layers
+4. https://github.com/ggerganov/llama.cpp/discussions/3489
+5. check lora example in examples
\ No newline at end of file
diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py
index 025405a2c..f42967f08 100755
--- a/convert-hf-to-gguf.py
+++ b/convert-hf-to-gguf.py
@@ -2782,7 +2782,7 @@ def parse_args() -> argparse.Namespace:
         help="model is executed on big endian machine",
     )
     parser.add_argument(
-        "model", type=Path,
+        "--model",dest="model", type=Path, default="models/phi-3-mini/",
         help="directory containing model file",
     )
     parser.add_argument(