Merge branch 'main' of github.com:abetlen/llama-cpp-python into main

abetlen · abetlen · commit bfb42b7d7fda · 2024-08-08T11:33:32.000-04:00
diff --git a/.github/workflows/build-and-release.yaml b/.github/workflows/build-and-release.yaml
@@ -29,7 +29,7 @@ jobs:
           python -m pip install -e .[all]
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.19.2
+        uses: pypa/cibuildwheel@v2.20.0
         env:
           # disable repair
           CIBW_REPAIR_WHEEL_COMMAND: ""
@@ -56,7 +56,7 @@ jobs:
           platforms: linux/arm64
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.19.2
+        uses: pypa/cibuildwheel@v2.20.0
         env:
           CIBW_SKIP: "*musllinux* pp*"
           CIBW_REPAIR_WHEEL_COMMAND: ""
diff --git a/.github/workflows/build-wheels-metal.yaml b/.github/workflows/build-wheels-metal.yaml
@@ -30,7 +30,7 @@ jobs:
           python -m pip install -e .[all]
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.19.2
+        uses: pypa/cibuildwheel@v2.20.0
         env:
           # disable repair
           CIBW_REPAIR_WHEEL_COMMAND: ""
diff --git a/llama_cpp/llama.py b/llama_cpp/llama.py
@@ -777,11 +777,12 @@ def generate(
                 else:
                     break
             if longest_prefix > 0:
-                if self.verbose:
-                    print("Llama.generate: prefix-match hit", file=sys.stderr)
                 reset = False
                 tokens = tokens[longest_prefix:]
                 self.n_tokens = longest_prefix
+                if self.verbose:
+                    print(f"Llama.generate: {longest_prefix} prefix-match hit, "
+                          f"remaining {len(tokens)} prompt tokens to eval", file=sys.stderr)                    
 
         # Reset the model state
         if reset:
@@ -2159,7 +2160,7 @@ def from_pretrained(
 
         files = [
             file["name"] if isinstance(file, dict) else file
-            for file in hffs.ls(repo_id)
+            for file in hffs.ls(repo_id, recursive=True))
         ]
 
         # split each file into repo_id, subfolder, filename
