Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,3 +1,40 @@
|
|
| 1 |
-
---
|
| 2 |
-
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model:
|
| 3 |
+
- Qwen/Qwen3-Coder-Next
|
| 4 |
+
tags:
|
| 5 |
+
- text-generation-inference
|
| 6 |
+
license: apache-2.0
|
| 7 |
+
---
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
**Qwen3-Coder-Next-REAP-48B-A3B** has the following specifications:
|
| 15 |
+
|
| 16 |
+
- **Type:** Causal Language Models
|
| 17 |
+
- **Number of Parameters**: 48B in total and 3B activated
|
| 18 |
+
- **Hidden Dimension**: 2048
|
| 19 |
+
- **Number of Layers**: 48
|
| 20 |
+
- **Hybrid Layout**: 12 * (3 * (Gated DeltaNet -> MoE) -> 1 * (Gated Attention -> MoE))
|
| 21 |
+
- **Gated Attention**:
|
| 22 |
+
- **Number of Attention Heads**: 16 for Q and 2 for KV
|
| 23 |
+
- **Head Dimension**: 256
|
| 24 |
+
- **Rotary Position Embedding Dimension**: 64
|
| 25 |
+
- **Gated DeltaNet**:
|
| 26 |
+
**Number of Linear Attention Heads: 32 for V and 16 for QK
|
| 27 |
+
**Head Dimension: 128
|
| 28 |
+
- **Mixture of Experts**:
|
| 29 |
+
- **Number of Experts: 308 (uniformly pruned from 512)
|
| 30 |
+
- **Number of Activated Experts: 10
|
| 31 |
+
- **Number of Shared Experts: 1
|
| 32 |
+
- **Context Length**: 262,144 natively
|
| 33 |
+
- **Compression Method**: REAP (Router-weighted Expert Activation Pruning)
|
| 34 |
+
- **Compression Ratio**: 40% expert pruning
|
| 35 |
+
|
| 36 |
+
Test video 1 (agentic task) @Q4_K_XL : https://www.bilibili.com/video/BV1f8cNzcEHV/
|
| 37 |
+
Prompt: please clone the repository https://github.com/ggml-org/llama.cpp in /home/lovedheart/llama_ and review the PR 19435.
|
| 38 |
+
|
| 39 |
+
Test video 2 -> fastllm (int8 quantization) approx. Q8_0 in GGUF : https://www.bilibili.com/video/BV1hwFJzXEVP/
|
| 40 |
+
Prompt: Create a cosmic nebula background using Three.js with the following requirements: a deep black space background with twinkling white stars; 2–3 large semi-transparent purple/pink nebula clouds with a smoky texture; slow rotation animation; optimized for white text display. Implementation details: 1. Starfield: 5000 white particles randomly distributed with subtle twinkling; 2. Nebula: 2–3 large purple particle clusters using additive blending mode; 3. Colors: #8B5CF6, #C084FC, #F472B6 (purple to pink gradient); 4. Animation: overall rotation.y += 0.001, stars' opacity flickering; 5. Setup: WebGLRenderer with alpha:true and black background.
|