Merge pull request #82 from rhymes-ai/vllm

fix: use offical vllm
rhymes-ai · Nov 27, 2024 · 7cbf499 · 7cbf499
2 parents 3233085 + 108b2af
commit 7cbf499
Show file tree

Hide file tree

Showing 4 changed files with 242 additions and 270 deletions.
diff --git a/inference/notebooks/01_single_image_understanding_vllm.ipynb b/inference/notebooks/01_single_image_understanding_vllm.ipynb
@@ -82,18 +82,8 @@
     "from PIL import Image\n",
     "\n",
     "from transformers import AutoTokenizer\n",
-    "from vllm import LLM, ModelRegistry, SamplingParams\n",
-    "from vllm.model_executor.models import _MULTIMODAL_MODELS\n",
+    "from vllm import LLM, SamplingParams\n",
     "\n",
-    "from aria.vllm.aria import AriaForConditionalGeneration\n",
-    "\n",
-    "ModelRegistry.register_model(\n",
-    "    \"AriaForConditionalGeneration\", AriaForConditionalGeneration\n",
-    ")\n",
-    "_MULTIMODAL_MODELS[\"AriaForConditionalGeneration\"] = (\n",
-    "    \"aria\",\n",
-    "    \"AriaForConditionalGeneration\",\n",
-    ")\n",
     "\n",
     "model_id_or_path = \"rhymes-ai/Aria\"\n",
     "\n",
@@ -574,9 +564,9 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "aria",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
-   "name": "aria"
+   "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {

diff --git a/inference/notebooks/02_multi_image_understanding_vllm.ipynb b/inference/notebooks/02_multi_image_understanding_vllm.ipynb
@@ -109,18 +109,8 @@
     "from PIL import Image\n",
     "\n",
     "from transformers import AutoTokenizer\n",
-    "from vllm import LLM, ModelRegistry, SamplingParams\n",
-    "from vllm.model_executor.models import _MULTIMODAL_MODELS\n",
+    "from vllm import LLM, SamplingParams\n",
     "\n",
-    "from aria.vllm.aria import AriaForConditionalGeneration\n",
-    "\n",
-    "ModelRegistry.register_model(\n",
-    "    \"AriaForConditionalGeneration\", AriaForConditionalGeneration\n",
-    ")\n",
-    "_MULTIMODAL_MODELS[\"AriaForConditionalGeneration\"] = (\n",
-    "    \"aria\",\n",
-    "    \"AriaForConditionalGeneration\",\n",
-    ")\n",
     "\n",
     "model_id_or_path = \"rhymes-ai/Aria\"\n",
     "\n",
@@ -847,9 +837,9 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "aria",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
-   "name": "aria"
+   "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {

diff --git a/inference/notebooks/03_multipage_pdf_understanding_vllm.ipynb b/inference/notebooks/03_multipage_pdf_understanding_vllm.ipynb
@@ -128,18 +128,8 @@
     "from PIL import Image\n",
     "\n",
     "from transformers import AutoTokenizer\n",
-    "from vllm import LLM, ModelRegistry, SamplingParams\n",
-    "from vllm.model_executor.models import _MULTIMODAL_MODELS\n",
+    "from vllm import LLM, SamplingParams\n",
     "\n",
-    "from aria.vllm.aria import AriaForConditionalGeneration\n",
-    "\n",
-    "ModelRegistry.register_model(\n",
-    "    \"AriaForConditionalGeneration\", AriaForConditionalGeneration\n",
-    ")\n",
-    "_MULTIMODAL_MODELS[\"AriaForConditionalGeneration\"] = (\n",
-    "    \"aria\",\n",
-    "    \"AriaForConditionalGeneration\",\n",
-    ")\n",
     "\n",
     "model_id_or_path = \"rhymes-ai/Aria\"\n",
     "\n",
@@ -714,9 +704,9 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "aria",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
-   "name": "aria"
+   "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {

diff --git a/inference/notebooks/04_video_understanding_vllm.ipynb b/inference/notebooks/04_video_understanding_vllm.ipynb