Style

notebooks/openvino/sentence_transformer_quantization.ipynb

@@ -177,9 +177,11 @@
     "\n",
     "quantizer = OVQuantizer.from_pretrained(model)\n",
     "\n",
+    "\n",
     "def preprocess_function(examples, tokenizer):\n",
     "    return tokenizer(examples[\"sentence\"], padding=\"max_length\", max_length=384, truncation=True)\n",
     "\n",
+    "\n",
     "calibration_dataset = quantizer.get_calibration_dataset(\n",
     "    \"glue\",\n",
     "    dataset_config_name=\"sst2\",\n",
@@ -194,13 +196,6 @@
     "tokenizer.save_pretrained(int8_ptq_model_path)"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
-  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -216,11 +211,12 @@
    "source": [
     "from transformers import Pipeline\n",
     "import torch.nn.functional as F\n",
-    "import torch \n",
+    "import torch\n",
+    "\n",
     "\n",
     "# copied from the model card\n",
     "def mean_pooling(model_output, attention_mask):\n",
-    "    token_embeddings = model_output[0] #First element of model_output contains all token embeddings\n",
+    "    token_embeddings = model_output[0]  # First element of model_output contains all token embeddings\n",
     "    input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()\n",
     "    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(input_mask_expanded.sum(1), min=1e-9)\n",
     "\n",
@@ -230,7 +226,7 @@
     "        # we don\"t have any hyperameters to sanitize\n",
     "        preprocess_kwargs = {}\n",
     "        return preprocess_kwargs, {}, {}\n",
-    "      \n",
+    "\n",
     "    def preprocess(self, inputs):\n",
     "        encoded_inputs = self.tokenizer(inputs, padding=True, truncation=True, return_tensors=\"pt\")\n",
     "        return encoded_inputs\n",
@@ -283,7 +279,7 @@
     "from datasets import load_dataset\n",
     "from evaluate import load\n",
     "\n",
-    "eval_dataset = load_dataset(\"glue\",\"stsb\",split=\"validation\")\n",
+    "eval_dataset = load_dataset(\"glue\", \"stsb\", split=\"validation\")\n",
     "metric = load(\"glue\", \"stsb\")"
    ]
   },
@@ -315,21 +311,22 @@
     }
    ],
    "source": [
-    "def compute_sentence_similarity(sentence_1, sentence_2,pipeline):\n",
+    "def compute_sentence_similarity(sentence_1, sentence_2, pipeline):\n",
     "    embedding_1 = pipeline(sentence_1)\n",
     "    embedding_2 = pipeline(sentence_2)\n",
     "    # compute cosine similarity between two sentences\n",
     "    return torch.nn.functional.cosine_similarity(embedding_1, embedding_2, dim=1)\n",
     "\n",
     "\n",
     "def evaluate_stsb(example):\n",
-    "  default = compute_sentence_similarity(example[\"sentence1\"], example[\"sentence2\"], vanilla_emb)\n",
-    "  quantized = compute_sentence_similarity(example[\"sentence1\"], example[\"sentence2\"], q8_emb)\n",
-    "  return {\n",
-    "      \"reference\": (example[\"label\"] - 1) / (5 - 1), # rescale to [0,1]\n",
-    "      \"default\": float(default),\n",
-    "      \"quantized\": float(quantized),\n",
-    "      }\n",
+    "    default = compute_sentence_similarity(example[\"sentence1\"], example[\"sentence2\"], vanilla_emb)\n",
+    "    quantized = compute_sentence_similarity(example[\"sentence1\"], example[\"sentence2\"], q8_emb)\n",
+    "    return {\n",
+    "        \"reference\": (example[\"label\"] - 1) / (5 - 1),  # rescale to [0,1]\n",
+    "        \"default\": float(default),\n",
+    "        \"quantized\": float(quantized),\n",
+    "    }\n",
+    "\n",
     "\n",
     "result = eval_dataset.map(evaluate_stsb)"
    ]
@@ -353,9 +350,13 @@
     "default_acc = metric.compute(predictions=result[\"default\"], references=result[\"reference\"])\n",
     "quantized = metric.compute(predictions=result[\"quantized\"], references=result[\"reference\"])\n",
     "\n",
-    "print(\"vanilla model: pearson=\", default_acc['pearson'])\n",
-    "print(\"quantized model: pearson=\", quantized['pearson'])\n",
-    "print(\"The quantized model achieves \", round(quantized[\"pearson\"]/default_acc[\"pearson\"],2)*100, \"% accuracy of the fp32 model\")"
+    "print(\"vanilla model: pearson=\", default_acc[\"pearson\"])\n",
+    "print(\"quantized model: pearson=\", quantized[\"pearson\"])\n",
+    "print(\n",
+    "    \"The quantized model achieves \",\n",
+    "    round(quantized[\"pearson\"] / default_acc[\"pearson\"], 2) * 100,\n",
+    "    \"% accuracy of the fp32 model\",\n",
+    ")"
    ]
   },
   {