Skip to content

Update sparse model SageMaker doc to support sparse embedding format #4323

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 21, 2025
+55 −9
Merged

Update sparse model SageMaker doc to support sparse embedding format #4323

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
64 changes: 55 additions & 9 deletions docs/model_serving_framework/deploy_sparse_model_to_SageMaker.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,7 +52,13 @@
"outputs": [],
"source": [
"%%writefile handler/code/requirements.txt\n",
"sentence-transformers==5.0.0"
"transformers==4.56.1\n",
"huggingface_hub==0.35.0\n",
"hf_xet==1.1.10\n",
"tokenizers==0.22.0\n",
"regex==2025.9.1\n",
"safetensors==0.6.2\n",
"sentence-transformers==5.1.0"
]
},
{
Expand Down Expand Up @@ -134,30 +140,64 @@
" )\n",
" print(f\"Using device: {self.device}\")\n",
" self.model = SparseEncoder(model_id, device=self.device, trust_remote_code=trust_remote_code)\n",
" self._warmup()\n",
" self.initialized = True\n",
"\n",
" def preprocess(self, requests):\n",
" def _warmup(self):\n",
" input_data = [{\"body\": [\"hello world\"] * 10}]\n",
" self.handle(input_data, None)\n",
"\n",
" def _preprocess(self, requests):\n",
" inputSentence = []\n",
" batch_idx = []\n",
" formats = [] # per-text format: \"word\" or \"token_id\"\n",
"\n",
" for request in requests:\n",
" request_body = request.get(\"body\")\n",
" if isinstance(request_body, bytearray):\n",
" request_body = request_body.decode(\"utf-8\")\n",
" request_body = json.loads((request_body))\n",
" if isinstance(request_body, list):\n",
"\n",
" # dict-based new schema: {\"texts\": str | list[str], \"sparse_embedding_format\": str}\n",
" if isinstance(request_body, dict):\n",
" texts = request_body.get(\"texts\")\n",
" fmt = request_body.get(\"sparse_embedding_format\", \"word\")\n",
" fmt = \"token_id\" if isinstance(fmt, str) and fmt.lower() == \"token_id\" else \"word\"\n",
"\n",
" if isinstance(texts, list):\n",
" inputSentence += texts\n",
" batch_idx.append(len(texts))\n",
" formats += [fmt] * len(texts)\n",
" else:\n",
" inputSentence.append(texts)\n",
" batch_idx.append(1)\n",
" formats.append(fmt)\n",
"\n",
" # legacy schemas\n",
" elif isinstance(request_body, list):\n",
" inputSentence += request_body\n",
" batch_idx.append(len(request_body))\n",
" formats += [\"word\"] * len(request_body)\n",
" else:\n",
" inputSentence.append(request_body)\n",
" batch_idx.append(1)\n",
" formats.append(\"word\")\n",
"\n",
" return inputSentence, batch_idx, formats\n",
"\n",
" return inputSentence, batch_idx\n",
" def _convert_token_ids(self, sparse_embedding):\n",
" token_ids = self.model.tokenizer.convert_tokens_to_ids([x[0] for x in sparse_embedding])\n",
" return [(str(token_ids[i]), sparse_embedding[i][1]) for i in range(len(token_ids))]\n",
"\n",
" def handle(self, data, context):\n",
" inputSentence, batch_idx = self.preprocess(data)\n",
" inputSentence, batch_idx, formats = self._preprocess(data)\n",
" model_output = self.model.encode_document(inputSentence, batch_size=max_bs)\n",
" sparse_embedding = list(map(dict,self.model.decode(model_output)))\n",
"\n",
" sparse_embedding_word = self.model.decode(model_output)\n",
" for i, fmt in enumerate(formats):\n",
" if fmt == \"token_id\":\n",
" sparse_embedding_word[i] = self._convert_token_ids(sparse_embedding_word[i])\n",
" sparse_embedding = list(map(dict, sparse_embedding_word))\n",
"\n",
" outputs = [sparse_embedding[s:e]\n",
" for s, e in zip([0]+list(itertools.accumulate(batch_idx))[:-1],\n",
Expand Down Expand Up @@ -424,8 +464,8 @@
"```json\n",
"POST /_plugins/_ml/connectors/_create\n",
"{\n",
" \"name\": \"test\",\n",
" \"description\": \"Test connector for Sagemaker model\",\n",
" \"name\": \"Sagemaker Connector: embedding\",\n",
" \"description\": \"The connector to sagemaker embedding model\",\n",
" \"version\": 1,\n",
" \"protocol\": \"aws_sigv4\",\n",
" \"credential\": {\n",
Expand All @@ -436,6 +476,7 @@
" \"region\": \"{region}\",\n",
" \"service_name\": \"sagemaker\",\n",
" \"input_docs_processed_step_size\": 2,\n",
" \"sparse_embedding_format\": \"word\"\n",
" },\n",
" \"actions\": [\n",
" {\n",
Expand All @@ -445,7 +486,12 @@
" \"content-type\": \"application/json\"\n",
" },\n",
" \"url\": \"https://runtime.sagemaker.{region}.amazonaws.com/endpoints/{predictor.endpoint_name}/invocations\",\n",
" \"request_body\": \"${parameters.input}\"\n",
" \"request_body\": \"\"\"\n",
" {\n",
" \"texts\": ${parameters.input},\n",
" \"sparse_embedding_format\": \"${parameters.sparse_embedding_format}\"\n",
" }\n",
" \"\"\"\n",
" }\n",
" ],\n",
" \"client_config\":{\n",
Expand Down
Loading