Initial commit

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+plutus_recommend_index.faiss filter=lfs diff=lfs merge=lfs -text
+tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text

Dockerfile

@@ -0,0 +1,29 @@
+
+FROM python:3.10-slim
+
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    gcc \
+    g++ \
+    wget \
+    curl \
+    git \
+    libfaiss-dev \
+    && rm -rf /var/lib/apt/lists/*
+
+
+WORKDIR /app
+
+
+COPY requirements.txt .
+
+
+RUN pip install --no-cache-dir -r requirements.txt
+
+
+COPY . .
+
+
+EXPOSE 7860
+
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "7860", "--workers", "2"]

app/main.py

@@ -0,0 +1,237 @@
+# app/main.py
+
+import os
+import json
+import logging
+from typing import Optional
+import asyncio
+import uvicorn
+from fastapi import FastAPI, HTTPException
+from fastapi.responses import JSONResponse, StreamingResponse
+from pydantic import BaseModel
+import torch
+
+from .model import PlutusModel, SummaryModel
+from .recommender import Recommender
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger("plutus.api")
+
+_CACHE_DIR = os.getenv("HF_HOME", "/home/user/app")
+
+DEFAULT_RECOMMEND_JSON = os.getenv(_CACHE_DIR, "recommend.json")
+RECOMMEND_INDEX_PATH = os.path.join(_CACHE_DIR, "plutus_recommend_index.faiss")
+RECOMMEND_META_PATH = os.path.join(_CACHE_DIR, "plutus_recommend_meta.json")
+
+
+class GenerateCache:
+    last_query: Optional[str] = None
+    last_topic: Optional[str] = None
+    last_personality: Optional[str] = None
+    last_level: Optional[str] = None
+    last_output: Optional[str] = None
+
+
+GEN_CACHE = GenerateCache()
+
+logger.info("Loading PlutusModel + Recommender...")
+
+plutus_model = PlutusModel()
+
+recommender = Recommender(
+    recommend_json_path=DEFAULT_RECOMMEND_JSON,
+    index_path=RECOMMEND_INDEX_PATH,
+    meta_path=RECOMMEND_META_PATH
+)
+
+summary_model_wrapper = SummaryModel(model_name="Qwen/Qwen3-4B-Instruct-2507")
+
+app = FastAPI(title="Plutus Learner API")
+
+
+
+
+class GenerateRequest(BaseModel):
+    personality: str
+    level: str
+    topic: str
+    query: str 
+    max_new_tokens: int = 700
+    temperature: float = 0.4
+    top_p: float = 0.5
+
+
+class RecommendRequest(BaseModel):
+    top_k: int = 5
+
+
+class SummaryRequest(BaseModel):
+    top_k: int = 5
+
+
+
+
+@app.get("/health")
+async def health():
+    return {"status": "ok", "device": plutus_model.device}
+
+
+
+
+
+@app.post("/generate")
+async def generate(req: GenerateRequest):
+
+  
+    prompt = plutus_model.create_prompt(
+        req.personality,
+        req.level,
+        req.topic,
+        req.query  
+    )
+
+    output_text = plutus_model.generate(
+        prompt,
+        max_new_tokens=req.max_new_tokens,
+        temperature=req.temperature,
+        top_p=req.top_p
+    )
+
+    # Cache everything for summary & follow-up queries
+    GEN_CACHE.last_query = req.query
+    GEN_CACHE.last_topic = req.topic
+    GEN_CACHE.last_personality = req.personality
+    GEN_CACHE.last_level = req.level
+    GEN_CACHE.last_output = output_text
+
+    return {
+        "text": output_text,
+        "topic": req.topic,
+        "query": req.query
+    }
+
+
+
+
+
+@app.get("/stream_generate")
+async def stream_generate(personality: str, level: str, topic: str):
+
+    prompt = plutus_model.create_prompt(personality, level, topic, "")
+
+    async def generate_events():
+        for chunk in plutus_model.stream_generate(prompt):
+            yield f"data: {json.dumps({'chunk': chunk})}\n\n"
+            await asyncio.sleep(0.01)
+
+    return StreamingResponse(generate_events(), media_type="text/event-stream")
+
+
+
+
+
+
+
+
+
+@app.post("/recommend")
+async def recommend(req: RecommendRequest):
+
+    if GEN_CACHE.last_query is None:
+        raise HTTPException(400, "No query found. Call /generate first.")
+
+    results = recommender.recommend_for_query(
+        query=GEN_CACHE.last_query,
+        top_k=req.top_k,
+        topic_boost=GEN_CACHE.last_topic
+    )
+
+    cleaned = [
+        {"topic": r["topic"], "type": r["type"], "url": r["url"]}
+        for r in results
+    ]
+
+    return {"query": GEN_CACHE.last_query, "results": cleaned}
+
+
+
+
+
+@app.post("/summary")
+async def summary(req: SummaryRequest):
+
+    if GEN_CACHE.last_output is None:
+        raise HTTPException(400, "No generate output found. Call /generate first.")
+
+    # Get recommended resources
+    recs = recommender.recommend_for_query(
+        query=GEN_CACHE.last_query,
+        top_k=req.top_k,
+        topic_boost=GEN_CACHE.last_topic
+    )
+
+    
+    pretty_recs = []
+    for r in recs:
+        pretty_recs.append(f"- ({r['type']}) {r['url']}")
+
+    formatted_resources_for_llm = "\n".join(pretty_recs)
+
+    
+    readable_resource_block = f"""
+Here are some helpful resources for further learning:
+{formatted_resources_for_llm}
+"""
+
+    
+    full_input_text = f"""
+Summarize the explanation below in a clear, simple, structured way.
+
+Your summary must include:
+1. A clean explanation of the topic
+2. A study roadmap
+3. A friendly explanation of the recommended learning resources (videos, docs)
+
+---- MAIN CONTENT ----
+{GEN_CACHE.last_output}
+
+---- RECOMMENDED RESOURCES ----
+{readable_resource_block}
+"""
+
+    # Include recommended=recs
+    summary_text = summary_model_wrapper.summarize_text(
+        full_input_text,
+        topic=GEN_CACHE.last_topic,
+        level=GEN_CACHE.last_level,
+        recommended=recs,
+        max_new_tokens=300
+    )
+
+ 
+    cleaned_resources = [
+        {"type": r["type"], "url": r["url"]}
+        for r in recs
+    ]
+
+    return {
+        "topic": GEN_CACHE.last_topic,
+        "summary": summary_text,
+        "resources": cleaned_resources
+    }
+
+
+
+
+
+
+
+@app.post("/admin/build_index")
+async def build_index(force: bool = False):
+    recommender.build_index(force=force)
+    return {"indexed": len(recommender.meta)}
+
+
+
+if __name__ == "__main__":
+    uvicorn.run("app.main:app", host="0.0.0.0", port=7860)

app/model.py

@@ -0,0 +1,287 @@
+import logging
+from threading import Thread
+from typing import Generator, Dict, Any, List
+import torch
+from transformers import (
+    AutoTokenizer,
+    AutoModelForCausalLM,
+    TextIteratorStreamer
+)
+
+logger = logging.getLogger("plutus.model")
+logging.basicConfig(level=logging.INFO)
+
+
+MAIN_MODEL_NAME = "Remostart/Plutus_Tutor_model"
+SUMMARY_MODEL_NAME = "Qwen/Qwen3-4B-Instruct-2507"
+
+
+
+class PlutusModel:
+    """
+    Handles the main learning model:
+    - Teaching prompt
+    - Synchronous generation
+    - Streaming generation
+    - Explaining recommendations
+    """
+
+    def __init__(self, model_name: str = MAIN_MODEL_NAME):
+        self.model_name = model_name
+        self.device = "cuda" if torch.cuda.is_available() else "cpu"
+
+        logger.info(f"[INIT] Main model running on: {self.device}")
+
+        self.tokenizer = None
+        self.model = None
+        self._load()
+
+
+    def _load(self):
+        """Loads the main teaching model and tokenizer."""
+        try:
+            logger.info(f"[LOAD] Loading tokenizer: {self.model_name}")
+            self.tokenizer = AutoTokenizer.from_pretrained(self.model_name, use_fast=True)
+
+            kwargs = {"torch_dtype": torch.float16} if self.device == "cuda" else {}
+
+            logger.info("[LOAD] Loading main model weights...")
+            self.model = AutoModelForCausalLM.from_pretrained(
+                self.model_name,
+                device_map="auto" if self.device == "cuda" else None,
+                low_cpu_mem_usage=True,
+                **kwargs
+            )
+
+            self.model.eval()
+            logger.info("[READY] Main model successfully loaded.")
+
+        except Exception as e:
+            logger.exception("Main model loading failed")
+            raise RuntimeError(f"Main model loading failed: {e}")
+
+
+   
+    def create_prompt(
+        self,
+        personality: str,
+        level: str,
+        topic: str,
+        extra_context: str = None
+    ) -> str:
+
+        prompt = (
+            f"You are PlutusTutor — the best expert in Cardano's Plutus smart contract ecosystem.\n\n"
+            f"User Info:\n"
+            f"- Personality: {personality}\n"
+            f"- Level: {level}\n"
+            f"- Topic: {topic}\n\n"
+            "Your task:\n"
+            "- Teach with extreme clarity.\n"
+            "- Give structured explanations.\n"
+            "- Include examples and, where needed, code.\n"
+            "- Avoid useless filler.\n"
+            "- Adapt tone slightly to user personality.\n\n"
+        )
+
+        if extra_context:
+            prompt += f"Additional Context:\n{extra_context}\n\n"
+
+        prompt += "Begin teaching now.\n\nAssistant:"
+
+        return prompt
+
+
+   
+    def generate(
+        self,
+        prompt: str,
+        max_new_tokens: int = 700,
+        temperature: float = 0.4,
+        top_p: float = 0.5
+    ) -> str:
+
+        try:
+            inputs = self.tokenizer(prompt, return_tensors="pt").to(self.model.device)
+
+            outputs = self.model.generate(
+                **inputs,
+                max_new_tokens=max_new_tokens,
+                do_sample=True,
+                temperature=temperature,
+                top_p=top_p,
+                eos_token_id=self.tokenizer.eos_token_id,
+                pad_token_id=self.tokenizer.pad_token_id
+            )
+
+            decoded = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+            if decoded.startswith(prompt):
+                decoded = decoded[len(prompt):].strip()
+
+            return decoded
+
+        except Exception as e:
+            logger.exception("Generation failed")
+            return f"[Generation Error] {e}"
+
+
+    def stream_generate(
+        self,
+        prompt: str,
+        max_new_tokens: int = 400,
+        temperature: float = 0.4,
+        top_p: float = 0.5
+    ) -> Generator[str, None, None]:
+
+        try:
+            inputs = self.tokenizer(prompt, return_tensors="pt").to(self.model.device)
+
+            streamer = TextIteratorStreamer(
+                self.tokenizer,
+                skip_prompt=True,
+                skip_special_tokens=True
+            )
+
+            thread = Thread(target=self.model.generate, kwargs={
+                **inputs,
+                "streamer": streamer,
+                "max_new_tokens": max_new_tokens,
+                "do_sample": True,
+                "temperature": temperature,
+                "top_p": top_p,
+                "eos_token_id": self.tokenizer.eos_token_id,
+                "pad_token_id": self.tokenizer.pad_token_id,
+            })
+            thread.start()
+
+            accumulated = ""
+            for chunk in streamer:
+                accumulated += chunk
+                yield accumulated
+
+        except Exception as e:
+            logger.exception("Streaming failed")
+            yield f"[Streaming Error] {e}"
+
+
+  
+    def summarize_recommendations(
+        self,
+        topic: str,
+        items: List[Dict[str, Any]],
+        personality: str = None,
+        level: str = None,
+        max_new_tokens: int = 120
+    ) -> str:
+
+        bullet_list = [
+            f"- {item['type'].upper()}: {item.get('title') or item.get('url')} ({item['url']})"
+            for item in items
+        ]
+
+        refs = "\n".join(bullet_list)
+
+        prompt = (
+            f"The user is learning: {topic}\n"
+            "Here are recommended videos and documents:\n\n"
+            f"{refs}\n\n"
+            "Explain why these choices are perfect for the user.\n"
+            f"Personality: {personality}\n"
+            f"Skill Level: {level}\n"
+            "Tone should be confident and friendly.\n\nAssistant:"
+        )
+
+        return self.generate(prompt, max_new_tokens=max_new_tokens)
+
+
+
+
+
+class SummaryModel:
+    """
+    Runs the summarization LLM:
+    - Summaries the full teaching text
+    - Adds clarity + structure
+    - Used in the /summary endpoint
+    """
+
+    def __init__(self, model_name: str = SUMMARY_MODEL_NAME):
+        self.model_name = model_name
+        self.device = "cuda" if torch.cuda.is_available() else "cpu"
+
+        logger.info(f"[INIT] Summary model running on: {self.device}")
+
+        self.tokenizer = None
+        self.model = None
+        self._load()
+
+
+    def _load(self):
+        try:
+            logger.info(f"[LOAD] Loading summary tokenizer: {self.model_name}")
+            self.tokenizer = AutoTokenizer.from_pretrained(self.model_name, use_fast=True)
+
+            kwargs = {"torch_dtype": torch.float16} if self.device == "cuda" else {}
+
+            logger.info(f"[LOAD] Loading summary model: {self.model_name}")
+            self.model = AutoModelForCausalLM.from_pretrained(
+                self.model_name,
+                device_map="auto" if self.device == "cuda" else None,
+                low_cpu_mem_usage=True,
+                **kwargs
+            )
+
+            self.model.eval()
+            logger.info("[READY] Summary model loaded.")
+
+        except Exception as e:
+            logger.exception("Summary model loading failed")
+            raise RuntimeError(f"Summary model loading failed: {e}")
+
+
+  
+    def summarize_text(
+        self,
+        full_teaching: str,
+        topic: str,
+        level: str,
+        recommended: List[Dict[str, Any]],
+        max_new_tokens: int = 400
+    ) -> str:
+
+        # Format RAG references
+        refs = "\n".join([
+            f"- {item['type'].upper()}: {item.get('title') or item.get('url')} ({item['url']})"
+            for item in recommended
+        ]) if recommended else "None"
+
+        prompt = (
+            f"You are a world-class summarization assistant.\n\n"
+            f"TOPIC: {topic}\n"
+            f"LEVEL: {level}\n\n"
+            f"Here is the full teaching content you must summarize:\n\n"
+            f"{full_teaching}\n\n"
+            "Now produce a clean, structured, extremely clear summary.\n"
+            "After the summary, recommend these resources clearly:\n\n"
+            f"{refs}\n\n"
+            "Assistant:"
+        )
+
+        inputs = self.tokenizer(prompt, return_tensors="pt").to(self.model.device)
+
+        out = self.model.generate(
+            **inputs,
+            max_new_tokens=max_new_tokens,
+            temperature=0.2,
+            top_p=0.85,
+            do_sample=True,
+            eos_token_id=self.tokenizer.eos_token_id
+        )
+
+        decoded = self.tokenizer.decode(out[0], skip_special_tokens=True)
+
+        if decoded.startswith(prompt):
+            decoded = decoded[len(prompt):].strip()
+
+        return decoded.strip()

app/recommender.py

@@ -0,0 +1,185 @@
+import os
+import json
+import logging
+from typing import List, Dict, Any
+import numpy as np
+
+logger = logging.getLogger("plutus.recommender")
+logging.basicConfig(level=logging.INFO)
+
+
+_EMB_MODEL_NAME = os.getenv("EMB_MODEL_NAME", "sentence-transformers/all-MiniLM-L6-v2")
+
+
+_CACHE_DIR = os.getenv("HF_HOME", "/home/user/app")
+_INDEX_FILE = os.path.join(_CACHE_DIR, "plutus_recommend_index.faiss")
+_META_FILE = os.path.join(_CACHE_DIR, "plutus_recommend_meta.json")
+
+try:
+    from sentence_transformers import SentenceTransformer
+    import faiss
+except Exception:
+    logger.warning(" sentence-transformers or faiss not installed. Ensure both are in requirements.txt")
+
+
+
+class Recommender:
+    """
+    Embedding-based semantic recommender for Plutus topics.
+    Loads resources from recommend.json, builds a FAISS index for fast similarity search.
+    """
+
+    def __init__(
+        self,
+        recommend_json_path: str,
+        emb_model_name: str = _EMB_MODEL_NAME,
+        index_path: str = _INDEX_FILE,
+        meta_path: str = _META_FILE,
+    ):
+        self.recommend_json_path = recommend_json_path
+        self.emb_model_name = emb_model_name
+        self.index_path = index_path
+        self.meta_path = meta_path
+        self.model = None
+        self.index = None
+        self.meta: List[Dict[str, Any]] = []
+        self.topics_map: Dict[str, Any] = {}
+
+      
+        self._load_json()
+
+        
+        self._maybe_init_embedding_model()
+
+        
+        if os.path.exists(self.index_path) and os.path.exists(self.meta_path):
+            try:
+                self._load_index()
+            except Exception:
+                logger.exception("Index load failed — will rebuild on demand.")
+        else:
+            logger.info("No index found — will build when first used.")
+
+
+    def _load_json(self):
+        """Load recommend.json file."""
+        if not os.path.exists(self.recommend_json_path):
+            raise FileNotFoundError(f"recommend.json not found at: {self.recommend_json_path}")
+        with open(self.recommend_json_path, "r", encoding="utf-8") as f:
+            self.topics_map = json.load(f)
+        logger.info(f"Loaded recommend.json with {len(self.topics_map)} topics.")
+
+    
+    def _maybe_init_embedding_model(self):
+        if self.model is None:
+            try:
+                from sentence_transformers import SentenceTransformer
+                self.model = SentenceTransformer(self.emb_model_name)
+                logger.info(f"Loaded embedding model: {self.emb_model_name}")
+            except Exception as e:
+                logger.exception(f" Failed to load embedding model: {e}")
+                raise RuntimeError("Embedding model not available. Please check dependencies.")
+
+  
+    def build_index(self, force: bool = False):
+        """
+        Builds FAISS index from recommend.json.
+        Each document and video becomes a searchable vector.
+        Automatically saves the index and metadata to disk.
+        """
+        if self.index is not None and not force:
+            logger.info("Index already built; skipping rebuild.")
+            return
+
+        items = []
+        texts = []
+
+        
+        for topic, val in self.topics_map.items():
+            for d in val.get("docs", []):
+                items.append({"topic": topic, "type": "doc", "url": d})
+                texts.append(f"{topic} doc {d}")
+            for v in val.get("videos", []):
+                items.append({"topic": topic, "type": "video", "url": v})
+                texts.append(f"{topic} video {v}")
+
+        if not texts:
+            raise ValueError("No docs/videos found in recommend.json to index.")
+
+        logger.info(f"Encoding {len(texts)} recommendation entries...")
+
+        emb = self.model.encode(texts, convert_to_numpy=True, show_progress_bar=False)
+        faiss.normalize_L2(emb)
+        d = emb.shape[1]
+
+        try:
+            index = faiss.IndexFlatIP(d)
+            index.add(emb)
+            self.index = index
+            self.meta = items
+            
+            try:
+                faiss.write_index(self.index, self.index_path)
+                with open(self.meta_path, "w", encoding="utf-8") as f:
+                    json.dump(self.meta, f, ensure_ascii=False, indent=2)
+                logger.info(f"Saved FAISS index and metadata ({len(items)} items).")
+            except Exception:
+                logger.warning(" Could not persist index — running in memory only (likely Hugging Face Space).")
+        except Exception as e:
+            logger.exception(f" Failed to build FAISS index: {e}")
+            raise RuntimeError(f"Index build failed: {e}")
+
+    
+    def _load_index(self):
+        """Loads index and metadata files."""
+        import faiss
+        self.index = faiss.read_index(self.index_path)
+        with open(self.meta_path, "r", encoding="utf-8") as f:
+            self.meta = json.load(f)
+        logger.info(f"Loaded FAISS index with {len(self.meta)} entries.")
+
+    
+    def recommend_for_query(self, query: str, top_k: int = 5, topic_boost: str = None) -> List[Dict[str, Any]]:
+        """
+        Returns top_k recommended items for `query`.
+        Uses cosine similarity (via normalized inner product).
+        """
+        if self.index is None:
+            logger.info("Index not found in memory — building now.")
+            self.build_index()
+
+        q_emb = self.model.encode([query], convert_to_numpy=True)
+        import faiss
+        faiss.normalize_L2(q_emb)
+
+        D, I = self.index.search(q_emb, top_k * 3)
+        results = []
+        seen = set()
+
+        for score, idx in zip(D[0], I[0]):
+            if idx < 0:
+                continue
+            meta = self.meta[idx]
+            key = (meta.get("url"), meta.get("type"))
+            if key in seen:
+                continue
+            seen.add(key)
+            results.append({
+                "topic": meta.get("topic"),
+                "type": meta.get("type"),
+                "url": meta.get("url"),
+                "score": float(score),
+            })
+            if len(results) >= top_k:
+                break
+
+        
+        if topic_boost:
+            results.sort(
+                key=lambda x: (0 if x["topic"].lower() == topic_boost.lower() else 1, -x["score"])
+            )
+        else:
+            results.sort(key=lambda x: -x["score"])
+
+        logger.info(f"Recommended {len(results)} items for query: '{query}'")
+        return results

plutus_recommend_meta.json

@@ -0,0 +1,377 @@
+[
+  {
+    "topic": "What is Plutus",
+    "type": "doc",
+    "url": "https://docs.cardano.org/developer-resources/smart-contracts/plutus"
+  },
+  {
+    "topic": "What is Plutus",
+    "type": "doc",
+    "url": "https://plutus-community.readthedocs.io/en/latest/"
+  },
+  {
+    "topic": "What is Plutus",
+    "type": "doc",
+    "url": "https://iohk.io/en/blog/posts/2021/04/13/plutus-what-you-need-to-know/"
+  },
+  {
+    "topic": "What is Plutus",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=igV7kMXcdpw"
+  },
+  {
+    "topic": "What is Plutus",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=SvaFFSqyVwM"
+  },
+  {
+    "topic": "What is Plutus",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=zUerLu_GOQs"
+  },
+  {
+    "topic": "Introduction to Plutus Core",
+    "type": "doc",
+    "url": "https://plutus.cardano.intersectmbo.org/docs/"
+  },
+  {
+    "topic": "Introduction to Plutus Core",
+    "type": "doc",
+    "url": "https://plutus.cardano.intersectmbo.org/docs/delve-deeper/languages"
+  },
+  {
+    "topic": "Introduction to Plutus Core",
+    "type": "doc",
+    "url": "https://github.com/IntersectMBO/plutus"
+  },
+  {
+    "topic": "Introduction to Plutus Core",
+    "type": "doc",
+    "url": "https://well-typed.com/blog/2022/08/plutus-cores/"
+  },
+  {
+    "topic": "Introduction to Plutus Core",
+    "type": "video",
+    "url": "https://blog.hachi.one/post/an-introduction-to-plutus-core/"
+  },
+  {
+    "topic": "Introduction to Haskell",
+    "type": "doc",
+    "url": "https://developers.cardano.org/docs/smart-contracts/plutus/"
+  },
+  {
+    "topic": "Introduction to Haskell",
+    "type": "doc",
+    "url": "https://developers.cardano.org/docs/get-started/haskell/onboarding/"
+  },
+  {
+    "topic": "Introduction to Haskell",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=igV7kMXcdpw"
+  },
+  {
+    "topic": "Introduction to Haskell",
+    "type": "video",
+    "url": "https://www.emurgo.io/press-news/the-basics-of-haskell-plutus-and-marlowe-cardano/"
+  },
+  {
+    "topic": "Introduction to Plutus Smart Contract",
+    "type": "doc",
+    "url": "https://plutus.cardano.intersectmbo.org/docs/category/example-an-auction-smart-contract"
+  },
+  {
+    "topic": "Introduction to Plutus Smart Contract",
+    "type": "doc",
+    "url": "https://intersectmbo.github.io/plutus-apps/main/"
+  },
+  {
+    "topic": "Introduction to Plutus Smart Contract",
+    "type": "doc",
+    "url": "https://www.joinplank.com/articles/cardano-e2e-dapp-tutorial"
+  },
+  {
+    "topic": "Introduction to Plutus Smart Contract",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=1vTsPQpCcTE"
+  },
+  {
+    "topic": "Introduction to Plutus Smart Contract",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=wNXKiQanLTc"
+  },
+  {
+    "topic": "Introduction to Plutus Smart Contract",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=Q2SaoISFgRE"
+  },
+  {
+    "topic": "Introduction to Plutus Smart Contract",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=-vdn7gDiT9s"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "doc",
+    "url": "https://plutus-pioneer-program.readthedocs.io/en/latest/pioneer/week1.html"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "doc",
+    "url": "https://arxiv.org/abs/2003.14271"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "doc",
+    "url": "https://www.joinplank.com/articles/debugging-plutus-an-introduction-to-low-level-cardano-transactions-in-the-alonzo-era"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "doc",
+    "url": "https://cardanofoundation.org/academy/video/cardano-eutxo-model"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=igV7kMXcdpw"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=Q2SaoISFgRE"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=xiNCkegNt0M"
+  },
+  {
+    "topic": "Introduction to eUTXO",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=BYT914XxqOQ"
+  },
+  {
+    "topic": "Introduction to Plutus Playground",
+    "type": "doc",
+    "url": "https://plutus-community.readthedocs.io/en/latest/"
+  },
+  {
+    "topic": "Introduction to Plutus Playground",
+    "type": "doc",
+    "url": "https://docs.cardano.org/developer-resources/smart-contracts/plutus"
+  },
+  {
+    "topic": "Introduction to Plutus Playground",
+    "type": "doc",
+    "url": "https://www.nmkr.io/step-by-step-guide/how-to-use-cardanos-plutus-playground"
+  },
+  {
+    "topic": "Introduction to Plutus Playground",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=HtjOWAEzWL8"
+  },
+  {
+    "topic": "Introduction to Plutus Playground",
+    "type": "video",
+    "url": "https://www.youtube.com/playlist?list=PL53JxaGwWUqAE59_XRxhwDuN1QrBgbIav"
+  },
+  {
+    "topic": "Introduction to Plutus Playground",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=DhRS-JvoCw8"
+  },
+  {
+    "topic": "Advantages of Plutus",
+    "type": "doc",
+    "url": "https://docs.cardano.org/developer-resources/smart-contracts/plutus"
+  },
+  {
+    "topic": "Advantages of Plutus",
+    "type": "doc",
+    "url": "https://www.essentialcardano.io/article/essential-resources-and-documentation-for-the-plutus-pioneer-program"
+  },
+  {
+    "topic": "Advantages of Plutus",
+    "type": "doc",
+    "url": "https://www.netsetsoftware.com/insights/mastering-cardano-smart-contracts-build-real-world-web-3-0-solutions-with-plutus/"
+  },
+  {
+    "topic": "Advantages of Plutus",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=SvaFFSqyVwM"
+  },
+  {
+    "topic": "Advantages of Plutus",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=oFgJg_DulKw"
+  },
+  {
+    "topic": "Advantages of Plutus",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=pJUg_2PI4xM"
+  },
+  {
+    "topic": "Advantages of Plutus",
+    "type": "video",
+    "url": "https://www.antiersolutions.com/blogs/cardano-plutus-smart-contract-a-detailed-discussion/"
+  },
+  {
+    "topic": "Versioning",
+    "type": "doc",
+    "url": "https://plutus.cardano.intersectmbo.org/docs/working-with-scripts/ledger-language-version"
+  },
+  {
+    "topic": "Versioning",
+    "type": "doc",
+    "url": "https://plutus.cardano.intersectmbo.org/docs/essential-concepts/versions/"
+  },
+  {
+    "topic": "Versioning",
+    "type": "doc",
+    "url": "https://medium.com/tap-in-with-taptools/plutus-v3-on-testnet-5130d1e4838a"
+  },
+  {
+    "topic": "Versioning",
+    "type": "doc",
+    "url": "https://cexplorer.io/article/it-is-smart-to-use-plutus-v2-applications"
+  },
+  {
+    "topic": "Versioning",
+    "type": "video",
+    "url": "https://plutus.cardano.intersectmbo.org/docs/delve-deeper/further-resources/videos"
+  },
+  {
+    "topic": "Versioning",
+    "type": "video",
+    "url": "https://www.youtube.com/channel/UCcAwSpbpQDDzEDRQqcDH8Iw/playlists"
+  },
+  {
+    "topic": "Versioning",
+    "type": "video",
+    "url": "https://iohk.io/en/blog/posts/2024/02/12/unlocking-more-opportunities-with-plutus-v3/"
+  },
+  {
+    "topic": "On Chain and Off-Chain",
+    "type": "doc",
+    "url": "https://docs.cardano.org/developer-resources/smart-contracts/plutus"
+  },
+  {
+    "topic": "On Chain and Off-Chain",
+    "type": "doc",
+    "url": "https://www.learningcardano.com/on-chain-validator-scripts/"
+  },
+  {
+    "topic": "On Chain and Off-Chain",
+    "type": "doc",
+    "url": "https://projectcatalyst.io/funds/5/developer-ecosystem/step-by-step-guide-off-chain-code"
+  },
+  {
+    "topic": "On Chain and Off-Chain",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=2MbzKzoBiak"
+  },
+  {
+    "topic": "On Chain and Off-Chain",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=NM8uqILIR0I"
+  },
+  {
+    "topic": "Cardano CLI",
+    "type": "doc",
+    "url": "https://plutus.cardano.intersectmbo.org/docs/using-plinth/cli-plutus"
+  },
+  {
+    "topic": "Cardano CLI",
+    "type": "doc",
+    "url": "https://hydra.family/head-protocol/docs/how-to/commit-script-utxo"
+  },
+  {
+    "topic": "Cardano CLI",
+    "type": "doc",
+    "url": "https://docs.cardano.org/developer-resources/transaction-tutorials"
+  },
+  {
+    "topic": "Cardano CLI",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=2MbzKzoBiak"
+  },
+  {
+    "topic": "Cardano CLI",
+    "type": "video",
+    "url": "https://aiken-lang.org/example--hello-world/end-to-end/cardano-cli"
+  },
+  {
+    "topic": "Data Types",
+    "type": "doc",
+    "url": "https://book.opshin.dev/smart_contract_tour/the_scriptcontext.html"
+  },
+  {
+    "topic": "Data Types",
+    "type": "video",
+    "url": "https://www.lidonation.com/en/posts/technical-overview-plutus-architecture"
+  },
+  {
+    "topic": "Monad",
+    "type": "doc",
+    "url": "https://www.youtube.com/watch?v=yOc1Z9zQ1zY"
+  },
+  {
+    "topic": "Monad",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=NM8uqILIR0I"
+  },
+  {
+    "topic": "Minting Policies",
+    "type": "doc",
+    "url": "https://meshjs.dev/apis/txbuilder/minting"
+  },
+  {
+    "topic": "Minting Policies",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=u1zFGKci3W4"
+  },
+  {
+    "topic": "State Machine",
+    "type": "doc",
+    "url": "https://learn.lovelace.academy/the-plutus-platform/state-machines/"
+  },
+  {
+    "topic": "State Machine",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=pO18jrw7GkA"
+  },
+  {
+    "topic": "Marlowe",
+    "type": "doc",
+    "url": "https://crypto.howtoclicks.com/2023/05/24/plutus-vs-marlowe/"
+  },
+  {
+    "topic": "Marlowe",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=X-bUcplFa0o"
+  },
+  {
+    "topic": "Cardano Tools (2025)",
+    "type": "doc",
+    "url": "https://cardano.org/developers/"
+  },
+  {
+    "topic": "Cardano Tools (2025)",
+    "type": "doc",
+    "url": "https://www.emurgo.io/press-news/here-are-5-useful-tools-for-cardano-developers/"
+  },
+  {
+    "topic": "Cardano Tools (2025)",
+    "type": "doc",
+    "url": "https://medium.com/tap-in-with-taptools/cardano-ecosystem-guide-q3-2025-af9d194020e2"
+  },
+  {
+    "topic": "Playgrounds - Local Setup",
+    "type": "doc",
+    "url": "https://plutuspbl.io/"
+  },
+  {
+    "topic": "Playgrounds - Local Setup",
+    "type": "video",
+    "url": "https://www.youtube.com/watch?v=g2F9raiGp_s"
+  }
+]

recommend.json

@@ -0,0 +1,202 @@
+{
+  "What is Plutus": {
+    "docs": [
+      "https://docs.cardano.org/developer-resources/smart-contracts/plutus",
+      "https://plutus-community.readthedocs.io/en/latest/",
+      "https://iohk.io/en/blog/posts/2021/04/13/plutus-what-you-need-to-know/"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=igV7kMXcdpw",
+      "https://www.youtube.com/watch?v=SvaFFSqyVwM",
+      "https://www.youtube.com/watch?v=zUerLu_GOQs"
+    ]
+  },
+  "Introduction to Plutus Core": {
+    "docs": [
+      "https://plutus.cardano.intersectmbo.org/docs/",
+      "https://plutus.cardano.intersectmbo.org/docs/delve-deeper/languages",
+      "https://github.com/IntersectMBO/plutus",
+      "https://well-typed.com/blog/2022/08/plutus-cores/"
+    ],
+    "videos": [
+      "https://blog.hachi.one/post/an-introduction-to-plutus-core/"
+    ]
+  },
+  "Introduction to Haskell": {
+    "docs": [
+      "https://developers.cardano.org/docs/smart-contracts/plutus/",
+      "https://developers.cardano.org/docs/get-started/haskell/onboarding/"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=igV7kMXcdpw",
+      "https://www.emurgo.io/press-news/the-basics-of-haskell-plutus-and-marlowe-cardano/"
+    ]
+  },
+  "Introduction to Plutus Smart Contract": {
+    "docs": [
+      "https://plutus.cardano.intersectmbo.org/docs/category/example-an-auction-smart-contract",
+      "https://intersectmbo.github.io/plutus-apps/main/",
+      "https://www.joinplank.com/articles/cardano-e2e-dapp-tutorial"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=1vTsPQpCcTE",
+      "https://www.youtube.com/watch?v=wNXKiQanLTc",
+      "https://www.youtube.com/watch?v=Q2SaoISFgRE",
+      "https://www.youtube.com/watch?v=-vdn7gDiT9s"
+    ]
+  },
+  "Introduction to eUTXO": {
+    "docs": [
+      "https://plutus-pioneer-program.readthedocs.io/en/latest/pioneer/week1.html",
+      "https://arxiv.org/abs/2003.14271",
+      "https://www.joinplank.com/articles/debugging-plutus-an-introduction-to-low-level-cardano-transactions-in-the-alonzo-era",
+      "https://cardanofoundation.org/academy/video/cardano-eutxo-model"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=igV7kMXcdpw",
+      "https://www.youtube.com/watch?v=Q2SaoISFgRE",
+      "https://www.youtube.com/watch?v=xiNCkegNt0M",
+      "https://www.youtube.com/watch?v=BYT914XxqOQ"
+    ]
+  },
+  "Introduction to Plutus Playground": {
+    "docs": [
+      "https://plutus-community.readthedocs.io/en/latest/",
+      "https://docs.cardano.org/developer-resources/smart-contracts/plutus",
+      "https://www.nmkr.io/step-by-step-guide/how-to-use-cardanos-plutus-playground"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=HtjOWAEzWL8",
+      "https://www.youtube.com/playlist?list=PL53JxaGwWUqAE59_XRxhwDuN1QrBgbIav",
+      "https://www.youtube.com/watch?v=DhRS-JvoCw8"
+    ]
+  },
+  "Advantages of Plutus": {
+    "docs": [
+      "https://docs.cardano.org/developer-resources/smart-contracts/plutus",
+      "https://www.essentialcardano.io/article/essential-resources-and-documentation-for-the-plutus-pioneer-program",
+      "https://www.netsetsoftware.com/insights/mastering-cardano-smart-contracts-build-real-world-web-3-0-solutions-with-plutus/"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=SvaFFSqyVwM",
+      "https://www.youtube.com/watch?v=oFgJg_DulKw",
+      "https://www.youtube.com/watch?v=pJUg_2PI4xM",
+      "https://www.antiersolutions.com/blogs/cardano-plutus-smart-contract-a-detailed-discussion/"
+    ]
+  },
+  "Versioning": {
+    "docs": [
+      "https://plutus.cardano.intersectmbo.org/docs/working-with-scripts/ledger-language-version",
+      "https://plutus.cardano.intersectmbo.org/docs/essential-concepts/versions/",
+      "https://medium.com/tap-in-with-taptools/plutus-v3-on-testnet-5130d1e4838a",
+      "https://cexplorer.io/article/it-is-smart-to-use-plutus-v2-applications"
+    ],
+    "videos": [
+      "https://plutus.cardano.intersectmbo.org/docs/delve-deeper/further-resources/videos",
+      "https://www.youtube.com/channel/UCcAwSpbpQDDzEDRQqcDH8Iw/playlists",
+      "https://iohk.io/en/blog/posts/2024/02/12/unlocking-more-opportunities-with-plutus-v3/"
+    ]
+  },
+  "Validation": {
+    "Low Level Validation": {
+      "docs": [
+        "https://iog-academy.gitbook.io/plutus-pioneers-program-fourth-cohort/lectures/lecture-02/lesson-01",
+        "https://plutus.cardano.intersectmbo.org/docs/essential-concepts/plinth-and-plutus-core",
+        "https://plutus-community.readthedocs.io/en/latest/#Plutus/Lectures/Cohort_03/Lecture_02/",
+        "https://iog-academy.gitbook.io/plutus-pioneers-program-fourth-cohort/lectures/lecture-02/lesson-01"
+      ],
+      "videos": [
+        "https://www.youtube.com/watch?v=2MbzKzoBiak",
+        "https://www.youtube.com/watch?v=3tcWCZV6L_w"
+      ]
+    },
+    "High Level Validation": {
+      "docs": [
+        "https://travishorn.github.io/ppp-notes/02-validation-scripts/03-high-level-typed-validation-scripts/",
+        "https://plutus.cardano.intersectmbo.org/docs/essential-concepts/plinth-and-plutus-core"
+      ],
+      "videos": [
+        "https://www.youtube.com/watch?v=2MbzKzoBiak",
+        "https://www.youtube.com/watch?v=NLZIrcXAlvU"
+      ]
+    }
+  },
+  "On Chain and Off-Chain": {
+    "docs": [
+      "https://docs.cardano.org/developer-resources/smart-contracts/plutus",
+      "https://www.learningcardano.com/on-chain-validator-scripts/",
+      "https://projectcatalyst.io/funds/5/developer-ecosystem/step-by-step-guide-off-chain-code"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=2MbzKzoBiak",
+      "https://www.youtube.com/watch?v=NM8uqILIR0I"
+    ]
+  },
+  "Cardano CLI": {
+    "docs": [
+      "https://plutus.cardano.intersectmbo.org/docs/using-plinth/cli-plutus",
+      "https://hydra.family/head-protocol/docs/how-to/commit-script-utxo",
+      "https://docs.cardano.org/developer-resources/transaction-tutorials"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=2MbzKzoBiak",
+      "https://aiken-lang.org/example--hello-world/end-to-end/cardano-cli"
+    ]
+  },
+  "Data Types": {
+    "docs": [
+      "https://book.opshin.dev/smart_contract_tour/the_scriptcontext.html"
+    ],
+    "videos": [
+      "https://www.lidonation.com/en/posts/technical-overview-plutus-architecture"
+    ]
+  },
+  "Monad": {
+    "docs": [
+      "https://www.youtube.com/watch?v=yOc1Z9zQ1zY"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=NM8uqILIR0I"
+    ]
+  },
+  "Minting Policies": {
+    "docs": [
+      "https://meshjs.dev/apis/txbuilder/minting"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=u1zFGKci3W4"
+    ]
+  },
+  "State Machine": {
+    "docs": [
+      "https://learn.lovelace.academy/the-plutus-platform/state-machines/"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=pO18jrw7GkA"
+    ]
+  },
+  "Marlowe": {
+    "docs": [
+      "https://crypto.howtoclicks.com/2023/05/24/plutus-vs-marlowe/"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=X-bUcplFa0o"
+    ]
+  },
+  "Cardano Tools (2025)": {
+    "docs": [
+      "https://cardano.org/developers/",
+      "https://www.emurgo.io/press-news/here-are-5-useful-tools-for-cardano-developers/",
+      "https://medium.com/tap-in-with-taptools/cardano-ecosystem-guide-q3-2025-af9d194020e2"
+    ],
+    "videos": []
+  },
+  "Playgrounds - Local Setup": {
+    "docs": [
+      "https://plutuspbl.io/"
+    ],
+    "videos": [
+      "https://www.youtube.com/watch?v=g2F9raiGp_s"
+    ]
+  }
+}

requirements.txt

@@ -0,0 +1,13 @@
+fastapi
+uvicorn
+transformers
+accelerate
+safetensors
+torch
+sentence-transformers
+faiss-cpu
+python-docx
+aiofiles
+python-multipart
+requests
+gunicorn

tokenizer/added_tokens.json

@@ -0,0 +1,28 @@
+{
+  "</think>": 151668,
+  "</tool_call>": 151658,
+  "</tool_response>": 151666,
+  "<think>": 151667,
+  "<tool_call>": 151657,
+  "<tool_response>": 151665,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}

tokenizer/chat_template.jinja

@@ -0,0 +1,61 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if message.content is string %}
+        {%- set content = message.content %}
+    {%- else %}
+        {%- set content = '' %}
+    {%- endif %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}

tokenizer/special_tokens_map.json

@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer/tokenizer_config.json

@@ -0,0 +1,239 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 1010000,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}