Update README.md

README.md

@@ -1,196 +1,238 @@
 ---
+license: apache-2.0
+base_model: Qwen/Qwen3.5-9B
 library_name: transformers
-tags: []
+pipeline_tag: image-text-to-text
+language:
+- en
+- hi
+- bn
+- mr
+- gu
+- te
+- ml
+- pa
+- kn
+- or
+- ta
+tags:
+- alpie
+- 169pi
+- qwen3.5
+- vision-language
+- multimodal
+- reasoning
+- coding
+- video-understanding
+- multilingual
+- indic
+- long-context
+- gguf
+- 9b
+- apache-2.0
 ---
 
-# Model Card for Model ID
+<div align="center">
 
-<!-- Provide a quick summary of what the model is/does. -->
+<h1>🌱 Alpie Seed 9B</h1>
 
+<p><b>A 9B open-weight vision-language model for reasoning, coding, and multimodal understanding.</b></p>
 
+<p>
+<a href="https://huggingface.co/169Pi/Alpie-Seed-9B"><img src="https://img.shields.io/badge/🤗_Model-Alpie_Seed_9B-FFD21E?style=for-the-badge&labelColor=1a1a2e" alt="Model"></a>
+<a href="https://huggingface.co/169Pi/Alpie-Seed-9B-GGUF"><img src="https://img.shields.io/badge/📦_GGUF-2/3/4--bit-7C3AED?style=for-the-badge&labelColor=1a1a2e" alt="GGUF"></a>
+<a href="https://ollama.com/169pi/alpie-core"><img src="https://img.shields.io/badge/🦙_Ollama-169pi/alpie--core-00C2A8?style=for-the-badge&labelColor=1a1a2e" alt="Ollama"></a>
+</p>
 
+<p>
+<img src="https://img.shields.io/badge/License-Apache_2.0-22C55E?style=flat-square" alt="license">
+<img src="https://img.shields.io/badge/Params-9B-6366F1?style=flat-square" alt="params">
+<img src="https://img.shields.io/badge/Context-262K-F59E0B?style=flat-square" alt="context">
+<img src="https://img.shields.io/badge/Modality-Text_·_Image_·_Video-EC4899?style=flat-square" alt="modality">
+<img src="https://img.shields.io/badge/Indic-10_languages-F97316?style=flat-square" alt="indic">
+<img src="https://img.shields.io/badge/Base-Qwen3.5--9B-0EA5E9?style=flat-square" alt="base">
+</p>
 
+<p>
+<a href="https://github.com/169Pi"><img src="https://img.shields.io/badge/GitHub-169Pi-181717?style=flat-square&logo=github" alt="GitHub"></a>
+<a href="https://www.linkedin.com/company/169pi"><img src="https://img.shields.io/badge/LinkedIn-169pi-0A66C2?style=flat-square&logo=linkedin&logoColor=white" alt="LinkedIn"></a>
+<a href="https://www.instagram.com/169pi.ai/"><img src="https://img.shields.io/badge/Instagram-169pi.ai-E4405F?style=flat-square&logo=instagram&logoColor=white" alt="Instagram"></a>
+<a href="https://www.youtube.com/@169pi"><img src="https://img.shields.io/badge/YouTube-@169pi-FF0000?style=flat-square&logo=youtube&logoColor=white" alt="YouTube"></a>
+<a href="https://x.com/169Pi_ai"><img src="https://img.shields.io/badge/X-@169Pi__ai-000000?style=flat-square&logo=x&logoColor=white" alt="X"></a>
+<a href="https://ollama.com/169pi/alpie-core"><img src="https://img.shields.io/badge/Ollama-169pi-FFFFFF?style=flat-square&logo=ollama&logoColor=black" alt="Ollama"></a>
+</p>
 
+</div>
 
-This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
-
-- **Developed by:** [More Information Needed]
-- **Funded by [optional]:** [More Information Needed]
-- **Shared by [optional]:** [More Information Needed]
-- **Model type:** [More Information Needed]
-- **Language(s) (NLP):** [More Information Needed]
-- **License:** [More Information Needed]
-- **Finetuned from model [optional]:** [More Information Needed]
-
-### Model Sources [optional]
-
-<!-- Provide the basic links for the model. -->
-
-- **Repository:** [More Information Needed]
-- **Paper [optional]:** [More Information Needed]
-- **Demo [optional]:** [More Information Needed]
-
-## Uses
-
-<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
-
-### Direct Use
-
-<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
-
-[More Information Needed]
-
-### Downstream Use [optional]
-
-<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
-
-[More Information Needed]
-
-### Out-of-Scope Use
-
-<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
-
-[More Information Needed]
-
-## Bias, Risks, and Limitations
-
-<!-- This section is meant to convey both technical and sociotechnical limitations. -->
-
-[More Information Needed]
-
-### Recommendations
-
-<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
-
-Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
-
-## How to Get Started with the Model
-
-Use the code below to get started with the model.
-
-[More Information Needed]
-
-## Training Details
-
-### Training Data
-
-<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
-
-[More Information Needed]
-
-### Training Procedure
-
-<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
-
-#### Preprocessing [optional]
-
-[More Information Needed]
-
-
-#### Training Hyperparameters
-
-- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
-
-#### Speeds, Sizes, Times [optional]
-
-<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
-
-[More Information Needed]
-
-## Evaluation
-
-<!-- This section describes the evaluation protocols and provides the results. -->
-
-### Testing Data, Factors & Metrics
-
-#### Testing Data
-
-<!-- This should link to a Dataset Card if possible. -->
-
-[More Information Needed]
-
-#### Factors
-
-<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
-
-[More Information Needed]
-
-#### Metrics
-
-<!-- These are the evaluation metrics being used, ideally with a description of why. -->
-
-[More Information Needed]
-
-### Results
-
-[More Information Needed]
-
-#### Summary
-
-
-
-## Model Examination [optional]
-
-<!-- Relevant interpretability work for the model goes here -->
-
-[More Information Needed]
-
-## Environmental Impact
-
-<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
-
-Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
-
-- **Hardware Type:** [More Information Needed]
-- **Hours used:** [More Information Needed]
-- **Cloud Provider:** [More Information Needed]
-- **Compute Region:** [More Information Needed]
-- **Carbon Emitted:** [More Information Needed]
-
-## Technical Specifications [optional]
-
-### Model Architecture and Objective
-
-[More Information Needed]
-
-### Compute Infrastructure
-
-[More Information Needed]
-
-#### Hardware
-
-[More Information Needed]
-
-#### Software
-
-[More Information Needed]
+---
 
-## Citation [optional]
+**Alpie Seed 9B** is a multimodal model developed by **169Pi AI**, built on the Qwen3.5-9B architecture and post-trained to strengthen reasoning, instruction following, coding, and visual understanding. It brings together text, image, and video understanding with a 262K context window and multiple quantized formats for deployment across different hardware.
+
+## Alpie Seed Highlights
+
+- **Unified vision-language model**: text, image, and video input on the Qwen3.5-9B architecture.
+- **Reasoning and coding**: post-trained with supervised fine-tuning and GRPO; supports thinking and non-thinking modes.
+- **262K native context** for long documents and extended conversations.
+- **Indic language coverage**: evaluated on MMLU and MILU across 10 Indic languages.
+- **Flexible deployment**: FP16 weights plus 2-bit, 3-bit, and 4-bit GGUF builds.
+- **Open license**: Apache 2.0.
+
+## Model Overview
+
+- Type: Causal Language Model with Vision Encoder
+- Training Stage: Post-training (SFT + GRPO) on Qwen3.5-9B
+- Language Model
+  - Number of Parameters: 9B
+  - Hidden Dimension: 4096
+  - Token Embedding: 248320 (Padded)
+  - Number of Layers: 32
+  - Hidden Layout: 8 × (3 × (Gated DeltaNet → FFN) → 1 × (Gated Attention → FFN))
+  - Gated DeltaNet:
+    - Number of Linear Attention Heads: 32 for V and 16 for QK
+    - Head Dimension: 128
+  - Gated Attention:
+    - Number of Attention Heads: 16 for Q and 4 for KV
+    - Head Dimension: 256
+    - Rotary Position Embedding Dimension: 64
+  - Feed Forward Network:
+    - Intermediate Dimension: 12288
+  - LM Output: 248320 (Padded)
+  - MTP: trained with multi-steps
+- Vision Encoder: 27-layer Vision Transformer
+- Context Length: 262,144 tokens
+
+The architecture is retained from Qwen3.5-9B; Alpie Seed's adaptation is carried out through post-training rather than a structural redesign.
+
+## Training
+
+Alpie Seed uses a two-stage post-training pipeline.
+
+1. **Supervised fine-tuning (SFT):** a LoRA adapter is trained to improve instruction following, response quality, and task performance.
+2. **Group Relative Policy Optimization (GRPO):** a trainable DoRA adapter is optimized while the SFT adapter is kept as a frozen reference policy. The GRPO stage uses a 4-bit NF4 base with double quantization and BF16 computation, and the trained adapter is merged into the base model at its original precision for release.
+
+Training used eight NVIDIA H100 GPUs, with 3,034 GRPO steps across five phases. The reward combines seven signals: answer correctness, hallucination penalty, self-consistency, educational quality, structure preservation, repetition penalty, and length penalty.
+
+<br>
+
+<div align="center">
+
+<img src="https://img.shields.io/badge/📊_BENCHMARK_RESULTS-7C3AED?style=for-the-badge" alt="📊_BENCHMARK_RESULTS">
+
+</div>
+
+## Benchmark Results
+
+<div align="center">
+
+![GPQA Diamond](https://img.shields.io/badge/GPQA_Diamond-84.32-F59E0B?style=for-the-badge) ![OCRBench](https://img.shields.io/badge/OCRBench-90.30-EC4899?style=for-the-badge) ![HallusionBench](https://img.shields.io/badge/HallusionBench-79.88-06B6D4?style=for-the-badge) ![RealWorldQA](https://img.shields.io/badge/RealWorldQA-80.39-22C55E?style=for-the-badge)
+
+</div>
+
+> Alpie Seed scores are from the FP16 model and do not apply to the 2-bit, 3-bit, or 4-bit variants. 🏆 gold badge = best score in the row; purple badge = Alpie Seed score. Comparison-model scores are taken from the [Qwen3.5-9B model card](https://huggingface.co/Qwen/Qwen3.5-9B).
+
+### 🧮 Language
+
+| | GPT-OSS-120B | GPT-OSS-20B | Qwen3-Next-80B-A3B-Thinking | Qwen3-30B-A3B-Thinking-2507 | Qwen3.5-9B | **Alpie Seed 9B** |
+|:---|:---:|:---:|:---:|:---:|:---:|:---:|
+| **Knowledge & STEM** | | | | | | |
+| MMLU | -- | -- | -- | -- | -- | ![87.30](https://img.shields.io/badge/87.30-7C3AED?style=flat-square) |
+| MMLU-Pro | 80.8 | 74.8 | **82.7** | 80.9 | 82.5 | ![82.54](https://img.shields.io/badge/82.54-7C3AED?style=flat-square) |
+| GPQA Diamond | 80.1 | 71.5 | 77.2 | 73.4 | 81.7 | ![84.32](https://img.shields.io/badge/84.32-F59E0B?style=flat-square) 🏆 |
+| TruthfulQA MC1 | -- | -- | -- | -- | -- | ![84.20](https://img.shields.io/badge/84.20-7C3AED?style=flat-square) |
+| TruthfulQA MC2 | -- | -- | -- | -- | -- | ![93.56](https://img.shields.io/badge/93.56-7C3AED?style=flat-square) |
+| **Mathematics** | | | | | | |
+| GSM8K | -- | -- | -- | -- | -- | ![98.00](https://img.shields.io/badge/98.00-7C3AED?style=flat-square) |
+| Math-500 | -- | -- | -- | -- | -- | ![91.00](https://img.shields.io/badge/91.00-7C3AED?style=flat-square) |
+| AIME 2026 | -- | -- | -- | -- | -- | ![83.33](https://img.shields.io/badge/83.33-7C3AED?style=flat-square) |
+| **Coding & Software Engineering** | | | | | | |
+| LiveCodeBench v6 | **82.7** | 74.6 | 68.7 | 66.0 | 65.6 | ![72.99](https://img.shields.io/badge/72.99-7C3AED?style=flat-square) |
+| HumanEval | -- | -- | -- | -- | -- | ![91.46](https://img.shields.io/badge/91.46-7C3AED?style=flat-square) |
+| HumanEval+ | -- | -- | -- | -- | -- | ![88.30](https://img.shields.io/badge/88.30-7C3AED?style=flat-square) |
+| MBPP | -- | -- | -- | -- | -- | ![84.30](https://img.shields.io/badge/84.30-7C3AED?style=flat-square) |
+| CRUXEval | -- | -- | -- | -- | -- | ![85.90](https://img.shields.io/badge/85.90-7C3AED?style=flat-square) |
+| SWE-Bench Verified | -- | -- | -- | -- | -- | ![61.20](https://img.shields.io/badge/61.20-7C3AED?style=flat-square) |
 
-<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+### 👁️ Vision Language
 
-**BibTeX:**
+| | GPT-5-Nano-2025-08-07 | Gemini-2.5-Flash-Lite | Qwen3-VL-30B-A3B | Qwen3.5-9B | **Alpie Seed 9B** |
+|:---|:---:|:---:|:---:|:---:|:---:|
+| **STEM and Puzzle** | | | | | |
+| MMMU-Pro | 57.2 | 59.7 | 63.0 | **70.1** | ![68.70](https://img.shields.io/badge/68.70-7C3AED?style=flat-square) |
+| MathVision | 62.2 | 52.1 | 65.7 | **78.9** | ![76.76](https://img.shields.io/badge/76.76-7C3AED?style=flat-square) |
+| **General VQA** | | | | | |
+| RealWorldQA | 71.8 | 72.2 | 77.4 | 80.3 | ![80.39](https://img.shields.io/badge/80.39-F59E0B?style=flat-square) 🏆 |
+| MMStar | 68.6 | 69.1 | 75.5 | **79.7** | ![78.87](https://img.shields.io/badge/78.87-7C3AED?style=flat-square) |
+| HallusionBench | 58.4 | 64.5 | 66.0 | 69.3 | ![79.88](https://img.shields.io/badge/79.88-F59E0B?style=flat-square) 🏆 |
+| **Text Recognition and Document Understanding** | | | | | |
+| OCRBench | 75.3 | 82.5 | 83.9 | 89.2 | ![90.30](https://img.shields.io/badge/90.30-F59E0B?style=flat-square) 🏆 |
+| DocVQA (Accuracy@0.5) | -- | -- | -- | -- | ![81.23](https://img.shields.io/badge/81.23-7C3AED?style=flat-square) |
+| DocVQA (Mean ANLS) | -- | -- | -- | -- | ![0.7981](https://img.shields.io/badge/0.7981-7C3AED?style=flat-square) |
+| TextVQA | -- | -- | -- | -- | ![78.67](https://img.shields.io/badge/78.67-7C3AED?style=flat-square) |
+| ChartQA | -- | -- | -- | -- | ![72.56](https://img.shields.io/badge/72.56-7C3AED?style=flat-square) |
+
+### 🇮🇳 Indic Language (MMLU & MILU)
+
+| Language | MMLU | MILU |
+|:---|:---:|:---:|
+| Hindi | ![77.55](https://img.shields.io/badge/77.55-F59E0B?style=flat-square) 🏆 | ![74.18](https://img.shields.io/badge/74.18-7C3AED?style=flat-square) |
+| Bengali | ![75.71](https://img.shields.io/badge/75.71-7C3AED?style=flat-square) | ![76.03](https://img.shields.io/badge/76.03-F59E0B?style=flat-square) 🏆 |
+| Marathi | ![75.22](https://img.shields.io/badge/75.22-7C3AED?style=flat-square) | ![70.36](https://img.shields.io/badge/70.36-7C3AED?style=flat-square) |
+| Gujarati | ![75.10](https://img.shields.io/badge/75.10-7C3AED?style=flat-square) | ![72.17](https://img.shields.io/badge/72.17-7C3AED?style=flat-square) |
+| Telugu | ![74.04](https://img.shields.io/badge/74.04-7C3AED?style=flat-square) | ![72.08](https://img.shields.io/badge/72.08-7C3AED?style=flat-square) |
+| Malayalam | ![72.97](https://img.shields.io/badge/72.97-7C3AED?style=flat-square) | ![68.22](https://img.shields.io/badge/68.22-7C3AED?style=flat-square) |
+| Punjabi | ![72.75](https://img.shields.io/badge/72.75-7C3AED?style=flat-square) | ![69.09](https://img.shields.io/badge/69.09-7C3AED?style=flat-square) |
+| Kannada | ![72.65](https://img.shields.io/badge/72.65-7C3AED?style=flat-square) | ![75.65](https://img.shields.io/badge/75.65-7C3AED?style=flat-square) |
+| Odia | ![72.63](https://img.shields.io/badge/72.63-7C3AED?style=flat-square) | ![70.28](https://img.shields.io/badge/70.28-7C3AED?style=flat-square) |
+| Tamil | ![68.10](https://img.shields.io/badge/68.10-7C3AED?style=flat-square) | ![71.23](https://img.shields.io/badge/71.23-7C3AED?style=flat-square) |
+| **Overall** | **73.67** | **73.01** |
 
-[More Information Needed]
+* SWE-Bench Verified: resolved rate using Mini-SWE-agent, with retries disabled.
+* Comparison-model scores were not re-run by 169Pi, so cross-model comparisons are indicative rather than strictly like-for-like. Scores vary with dataset versions, prompts, evaluation harnesses, and inference settings.
+* Empty cells (--) indicate scores not yet available or not applicable.
 
-**APA:**
+## Quickstart
 
-[More Information Needed]
+Serve with vLLM (OpenAI-compatible API):
 
-## Glossary [optional]
+```shell
+vllm serve 169Pi/Alpie-Seed-9B --port 8000 --max-model-len 262144 --reasoning-parser qwen3
+```
 
-<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+Run locally with Ollama:
 
-[More Information Needed]
+```shell
+ollama run 169pi/alpie-core
+```
 
-## More Information [optional]
+Quantized 2-bit, 3-bit, and 4-bit builds are in the [GGUF repository](https://huggingface.co/169Pi/Alpie-Seed-9B-GGUF). Thinking and non-thinking behavior is controlled through the chat template.
 
-[More Information Needed]
+## Intended Use and Limitations
 
-## Model Card Authors [optional]
+Alpie Seed is intended for research, development, and general-purpose applications: reasoning, mathematics, coding, image/chart/document understanding, multilingual use, long-context analysis, and local deployments.
 
-[More Information Needed]
+- Outputs may contain factual errors, unsupported conclusions, or hallucinations.
+- Performance varies across languages, domains, and task complexity.
+- Long context does not guarantee perfect recall across the whole window.
+- Quantization may affect quality, speed, and memory differently across hardware.
+- Not a substitute for professional judgment; do not use as the sole basis for medical, legal, financial, or other high-impact decisions.
 
-## Model Card Contact
+## Citation
 
-[More Information Needed]
+```bibtex
+@misc{alpie_seed_9b_2026,
+  title        = {Alpie Seed 9B},
+  author       = {{169Pi AI}},
+  year         = {2026},
+  howpublished = {\url{https://huggingface.co/169Pi/Alpie-Seed-9B}}
+}
+```
+
+## License
+
+Apache License 2.0. Users are responsible for ensuring their use and deployment comply with applicable laws and regulations.
+
+<div align="center">
+
+**Developed by 169Pi AI** 🌱 · [GitHub](https://github.com/169Pi) · [LinkedIn](https://www.linkedin.com/company/169pi) · [Instagram](https://www.instagram.com/169pi.ai/) · [YouTube](https://www.youtube.com/@169pi) · [X](https://x.com/169Pi_ai) · [Ollama](https://ollama.com/169pi/alpie-core)
+
+</div>