Update README.md

README.md

@@ -4,62 +4,64 @@ license_name: prism-research
 license_link: LICENSE.md
 language:
 - en
-- cn
+- zh
 tags:
 - glm4
 - prism
+- moe
 pipeline_tag: text-generation
 library_name: transformers
 ---
 
-[![Model](https://img.shields.io/badge/Model-2.6B-blue)]()
-[![Architecture](https://img.shields.io/badge/Architecture-LFM2%20Hybrid-green)]()
+[![Parameters](https://img.shields.io/badge/Parameters-30B--A3B_MoE-blue)]()
+[![Architecture](https://img.shields.io/badge/Architecture-GLM--4-green)]()
 [![Context](https://img.shields.io/badge/Context-128K-orange)]()
 
+# GLM-4.7-Flash-PRISM
 
+An unrestricted version of ZAI's GLM-4.7-Flash with over-refusal mechanisms removed using PRISM (Projected Refusal Isolation via Subspace Modification).
 
+<div align="center">
 
-## Model Description
+### ☕ Support Our Work
 
-This is **Ex0bit/GLM-4.7-Flash-PRISM**
+If you find this useful, consider supporting us on Ko-fi!
 
-<div align="center" style="background:#161b22;border:1px solid #30363d;border-radius:10px;padding:20px 24px;margin:20px 0;">
+[![Ko-fi](https://img.shields.io/badge/Ko--fi-Support%20Us-ff5e5b?logo=ko-fi&logoColor=white)](https://ko-fi.com/ericelbaz)
 
+| Option | Description |
+|--------|-------------|
+| [**PRISM VIP Membership**](https://ko-fi.com/summary/6bae206c-a751-4868-8dc7-f531afd1fb4c) | Access to all PRISM models |
+| [**One-Time Support**](https://ko-fi.com/s/86882e8991) | Support this model |
 
-<span style="color:#58a6ff;font-size:11px;font-weight:600;letter-spacing:0.5px;">PLEASE SUPPORT OUR WORK!</span>
-If you enjoy what we do, consider supporting us on Ko-fi! Every little bit means the world! https://ko-fi.com/ericelbaz
-<div align="center">
-  <a href="https://ko-fi.com/ericelbaz#tier17681523526070">
-    <img src="https://cdn-uploads.huggingface.co/production/uploads/63adf1fa42fd3b8dbaeb0c92/Qe17Xd59xWbucl1ZOl1jF.png" width="50%"/>
-  </a>
+</div>
 
-<br><br>
-<!--<span style="color:#e6edf3;font-size:18px;font-weight:600;">Complete both steps.</span>-->
-<br><br>
+---
 
-<table cellpadding="0" cellspacing="0" border="0" align="center" style="text-align:left;">
-  <!--<tr><td style="padding:6px 0;color:#e6edf3;font-size:13px;"><span style="color:#58a6ff;font-weight:600;margin-right:10px;">①</span>Submit the access request form below</td></tr>-->
-  <tr><td style="padding:6px 0;color:#e6edf3;font-size:13px;"><span style="color:#58a6ff;font-weight:600;margin-right:10px;">②</span>Support Donation Option:</td></tr>
-</table>
-<br>
+## Model Highlights
 
-<a href="https://ko-fi.com/summary/6bae206c-a751-4868-8dc7-f531afd1fb4c" target="_blank" style="background:#238636;color:white;text-decoration:none;padding:8px 16px;border-radius:6px;font-weight:600;font-size:16px;">PRISM VIP Member Sign-Up </a> <span style="color:#8b949e;font-size:11px;margin-left:8px;">  All Models </span>
-<br><br>
-<a href="https://ko-fi.com/s/86882e8991" target="_blank" style="background:#21262d;color:#e6edf3;text-decoration:none;padding:8px 16px;border-radius:6px;font-weight:500;font-size:16px;border:1px solid #30363d;">One-Time Support </a> <span style="color:#8b949e;font-size:11px;margin-left:8px;"> This Model </span>
-<br><br>
+- **PRISM Ablation** — State-of-the-art technique that removes over-refusal behaviors while preserving model capabilities
+- **30B-A3B MoE Architecture** — 30 billion total parameters with ~3 billion active per token for fast, efficient inference
+- **128K Context Window** — Extended context for complex tasks and large codebases
+- **Interleaved Thinking** — Multi-turn reasoning that persists across conversations with per-turn thinking control
 
-<span style="color:#3fb950;font-size:11px;">✓ Priority Access</span>
+## Benchmarks
 
-</div>
+| Benchmark | Score |
+|-----------|-------|
+| AIME 2025 | 91.6% |
+| τ²-Bench | 79.5% |
+| SWE-bench Verified | 59.2% |
+| GPQA | 75.2% |
 
-**GLM-4.7-Flash-PRISM:** Unrestricted (Zero Over-Refusals and Zero Propoganda) GLM-4.7-Flash Model Access
+## Usage
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-Access GLM-4.7-Flash-PRISM, a PRISM unchained version of ZAI's efficient 30B-A3B MoE model with over-refusal mechanisms removed.
+model = AutoModelForCausalLM.from_pretrained("Ex0bit/GLM-4.7-Flash-PRISM")
+tokenizer = AutoTokenizer.from_pretrained("Ex0bit/GLM-4.7-Flash-PRISM")
+```
 
-**What You Get:**
+## License
 
-- **PRISM (Projected Refusal Isolation via Subspace Modification)** — State-of-the-art abliteration technique that removes over-refusal behaviors while preserving capabilities
-- **30B-A3B MoE Architecture** — Lightweight yet powerful Mixture-of-Experts model with 30 billion total parameters and ~3 billion active per token for fast, efficient inference
-- **128K Context Window** — Extended context for complex tasks and large codebases
-- **Interleaved & Preserved Thinking** — Multi-turn reasoning that persists across conversations with per-turn thinking control
-- **Strong In-Class Benchmarks** — 91.6% AIME 2025, 79.5% τ²-Bench, 59.2% SWE-bench Verified, 75.2% GPQA
+This model is released under the [PRISM Research License](LICENSE.md).