File size: 3,637 Bytes
c9ac977
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8cbd715
 
c9ac977
 
57ba57e
c9ac977
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
---
license: apache-2.0
---
# HRWKV7-hxa079-Qwen3-14B

### Model Description

HRWKV7-hxa079-Qwen3-14B is an experimental hybrid architecture model that combines RWKV v7's linear attention mechanism with Group Query Attention (GQA) layers. Built upon the Qwen3-14B Instruct foundation, this model replaces most Transformer attention blocks with RWKV blocks while strategically maintaining some GQA layers to enhance performance on specific tasks.

- **Developed by:** OpenMOSE
- **Model type:** Hybrid Linear-Attention Language Model
- **Language(s):** Multilingual (inherited from Qwen3-14B)
- **License:** Apache-2.0
- **Base Model:** Qwen3-14B Instruct
- **Year:** 2025

### Architecture Specifications

- **Architecture:** RWKV v7 based "hxa079" Architecture + Group Query Attention Hybrid
- **Total Layers:** 40 layers (L40D5120)
  - 34 RWKV layers (with Rope)
  - 6 GQA layers (No Rope, No Position Embeddings)
- **Hidden Dimension:** 5120
- **Training Context Window:** 4096 tokens
- **Inference Context Window** 16384 tokens(100% NIAH)

## Technical Innovation

### RWKV "hxa079" Architecture

The model implements several key improvements over standard RWKV architectures:

1. **Token Shift Removal**: Unlike traditional RWKV, the hxa079 variant removes token shifting mechanisms
2. **GroupNorm Removal**: Eliminates GroupNorm layers for training stability
3. **k_first Introduction**: Implements a novel k_first mechanism optimized for attention conversion

### Hybrid Design Benefits

- **Linear Attention Inference**: RWKV blocks enable O(1) memory complexity during inference
- **Enhanced Needle Tasks**: Strategic placement of GQA layers significantly improves performance on needle-in-haystack retrieval tasks, addressing a known limitation of pure linear attention models
- **Implicit Position Encoding**: Interestingly, the model achieves better performance when RoPE (Rotary Position Embedding) is not applied to GQA layers, suggesting that RWKV blocks provide implicit positional encoding capabilities

## Intended Use

This is an **experimental research model** designed to explore hybrid architectures combining linear and quadratic attention mechanisms. It is intended for:

- Research into efficient attention mechanisms
- Benchmarking hybrid architecture performance
- Exploring linear attention limitations and solutions
- Academic and industrial R&D purposes

## Limitations

- **Experimental Status**: This model is in experimental stages and may exhibit unexpected behaviors
- **Context Window**: Limited to 4096 tokens during training, though RWKV architecture theoretically supports longer sequences
- **Performance Variability**: As a hybrid model, performance may vary significantly across different task types

## Training Details

- **Training Context Window:** 4096 tokens
- **Base Model Initialization:** Weights initialized from Qwen3-14B Instruct
- **Architecture Conversion:** Transformer attention blocks systematically replaced with RWKV blocks, except for 6 strategically placed GQA layers

## Evaluation

Performance evaluation is ongoing. The model shows promising results in:
- Maintaining base model capabilities while achieving linear attention efficiency
- Significantly improved needle-in-haystack task performance compared to pure RWKV architectures
- Competitive performance on standard language modeling benchmarks


## Model Card Contact

OpenMOSE - 2025

---

*Note: This is an experimental model. Performance characteristics and behaviors may differ from both pure RWKV and standard Transformer architectures. Users should thoroughly evaluate the model for their specific use cases.*