Add YOLOv8n weights

SA_Model/config.json

@@ -0,0 +1,33 @@
+{
+  "activation": "gelu",
+  "architectures": [
+    "DistilBertForSequenceClassification"
+  ],
+  "attention_dropout": 0.1,
+  "dim": 768,
+  "dropout": 0.1,
+  "dtype": "float32",
+  "finetuning_task": "sst-2",
+  "hidden_dim": 3072,
+  "id2label": {
+    "0": "NEGATIVE",
+    "1": "POSITIVE"
+  },
+  "initializer_range": 0.02,
+  "label2id": {
+    "NEGATIVE": 0,
+    "POSITIVE": 1
+  },
+  "max_position_embeddings": 512,
+  "model_type": "distilbert",
+  "n_heads": 12,
+  "n_layers": 6,
+  "output_past": true,
+  "pad_token_id": 0,
+  "qa_dropout": 0.1,
+  "seq_classif_dropout": 0.2,
+  "sinusoidal_pos_embds": false,
+  "tie_weights_": true,
+  "transformers_version": "4.56.1",
+  "vocab_size": 30522
+}

SA_Model/handler.py

@@ -0,0 +1,28 @@
+# handler.py
+import time
+from typing import Any, Dict, List, Union
+from transformers import pipeline
+
+class EndpointHandler:
+    def __init__(self, path: str = ""):
+        # Load a standard text-classification pipeline from local repo files
+        self.pipe = pipeline("text-classification", model=path)
+
+    def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
+        # Accept {"inputs": "..."} or {"inputs": ["...", "..."]}
+        inputs: Union[str, List[str]] = data.get("inputs", data)
+
+        t0 = time.perf_counter()
+        preds = self.pipe(inputs, truncation=True)
+        elapsed = time.perf_counter() - t0
+
+        # Normalize to list
+        preds_list = preds if isinstance(preds, list) else [preds]
+        labels = [p["label"] for p in preds_list]
+        scores = [float(p["score"]) for p in preds_list]
+
+        return {
+            "labels": labels,
+            "scores": scores,
+            "processing_time_sec": elapsed,
+        }

SA_Model/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61afb985df6fbe77fd84138129481f0a4444c02893782984e330f874f43cee4b
+size 267832560

SA_Model/prepare_model.py

@@ -0,0 +1,11 @@
+# prepare_model.py
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+MODEL = "distilbert-base-uncased-finetuned-sst-2-english"
+
+m = AutoModelForSequenceClassification.from_pretrained(MODEL)
+t = AutoTokenizer.from_pretrained(MODEL)
+
+m.save_pretrained(".")      # saves config.json + weights (bin/safetensors)
+t.save_pretrained(".")      # saves tokenizer files
+print("Saved model + tokenizer to repo root")

SA_Model/readme.md

@@ -0,0 +1,28 @@
+---
+tags:
+- sentiment-analysis
+- text-classification
+pipeline_tag: text-classification
+language:
+- en
+license: apache-2.0
+library_name: transformers
+model_name: distilbert-base-uncased-finetuned-sst-2-english
+---
+
+# Ericsson Day Demo Sentiment Model
+
+This model is based on **DistilBERT fine-tuned on SST-2** for sentiment analysis.  
+It has been wrapped with a custom `model.py` that also reports the internal processing time.
+
+## Model description
+- **Task**: Sentiment classification (positive/negative)
+- **Base model**: `distilbert-base-uncased-finetuned-sst-2-english`
+- **Wrapper**: Custom `model.py` that returns `"processing_time_sec"`
+
+## Example usage
+
+```python
+from transformers import pipeline
+classifier = pipeline("sentiment-analysis", model="ED-Demo/Ericsson_day_demo_model")
+print(classifier("I love Hugging Face!"))

SA_Model/requirements.txt

@@ -0,0 +1,2 @@
+torch
+transformers

SA_Model/special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "[CLS]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

SA_Model/tokenizer_config.json

@@ -0,0 +1,58 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "100": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "101": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "102": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "103": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": true,
+  "extra_special_tokens": {},
+  "mask_token": "[MASK]",
+  "model_max_length": 512,
+  "never_split": null,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "strip_accents": null,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "DistilBertTokenizer",
+  "unk_token": "[UNK]"
+}

Yolo_Model/.gitattributes

@@ -0,0 +1 @@
+Yolo_Model/yolov8n.pt filter=lfs diff=lfs merge=lfs -text

tests/test_SA_RTT.py

@@ -0,0 +1,33 @@
+import os
+import time
+import requests
+import statistics
+
+# Your endpoint + token
+ENDPOINT_URL = "https://igyvdwvwkcoglpbx.eu-west-1.aws.endpoints.huggingface.cloud"
+HF_TOKEN = os.getenv("HF_TOKEN")  # put your token in environment: export HF_TOKEN=hf_xxx
+
+headers = {
+    "Authorization": f"Bearer {HF_TOKEN}",
+    "Content-Type": "application/json"
+}
+
+payload = {"inputs": "This is fantastic!"}
+
+latencies = []
+
+# Warmup (optional: remove if you want cold start time)
+requests.post(ENDPOINT_URL, headers=headers, json=payload)
+
+for i in range(50):
+    t0 = time.perf_counter()
+    r = requests.post(ENDPOINT_URL, headers=headers, json=payload)
+    dt = time.perf_counter() - t0
+    latencies.append(dt)
+    print(f"Run {i+1}: RTT={dt:.3f}s, Response={r.json()}")
+
+print("\n📊 RTT stats (10 runs)")
+print(f"  Min   : {min(latencies):.3f}s")
+print(f"  Avg   : {statistics.mean(latencies):.3f}s")
+print(f"  Median: {statistics.median(latencies):.3f}s")
+print(f"  Max   : {max(latencies):.3f}s")

tests/test_roundtrip.py

@@ -1,24 +0,0 @@
-import time, statistics, os
-from huggingface_hub import InferenceClient
-
-ENDPOINT_URL = "https://<your-endpoint>.endpoints.huggingface.cloud"
-HF_TOKEN = os.getenv("HF_TOKEN")
-
-client = InferenceClient(endpoint_url=ENDPOINT_URL, token=HF_TOKEN)
-
-def measure_rtt(n=5):
-    print("⚡ Warming up...")
-    client.post(json={"inputs": "Warmup test"})
-
-    latencies = []
-    for i in range(n):
-        t0 = time.perf_counter()
-        out = client.post(json={"inputs": "This is fantastic!"})
-        dt = time.perf_counter() - t0
-        latencies.append(dt)
-        print(f"Run {i+1}: RTT={dt:.3f}s | Output={out}")
-
-    print(f"\n✅ RTT stats: min={min(latencies):.3f}s, median={statistics.median(latencies):.3f}s, max={max(latencies):.3f}s")
-
-if __name__ == "__main__":
-    measure_rtt()

tests/test_yolo.py

@@ -0,0 +1,52 @@
+import cv2
+from ultralytics import YOLO
+
+# Load model
+model = YOLO("yolov8n.pt")  # or path to your custom weights
+
+# Force index 0 (usually the MacBook’s built-in camera) using AVFoundation
+cap = cv2.VideoCapture(2, cv2.CAP_AVFOUNDATION)
+
+if not cap.isOpened():
+    print("Cannot open camera 0, try index 1")
+    cap = cv2.VideoCapture(1, cv2.CAP_AVFOUNDATION)
+    if not cap.isOpened():
+        raise IOError("Cannot open any webcam")
+
+while True:
+    ret, frame = cap.read()
+    if not ret:
+        break
+
+    # Run YOLO inference
+    results = model(frame)
+
+    # Draw results on frame
+    annotated_frame = results[0].plot()
+
+    # Get inference time (ms)
+    inf_time = results[0].speed["inference"]
+    text = f"Inference: {inf_time:.1f} ms"
+
+    # Overlay text on the frame
+    cv2.putText(
+        annotated_frame,
+        text,
+        (20, 40),  # position (x, y)
+        cv2.FONT_HERSHEY_SIMPLEX,
+        1,         # font scale
+        (0, 255, 0),  # green color
+        2,         # thickness
+        cv2.LINE_AA
+    )
+
+    # Show
+    cv2.imshow("YOLOv8 Webcam", annotated_frame)
+
+    # Press 'q' to quit
+    key = cv2.waitKey(10) & 0xFF
+    if key == ord("q"):
+        break
+
+cap.release()
+cv2.destroyAllWindows()