Card: consistent layout, file table, usage and metadata
Browse filesREADME.md only. Same licences and credits as before; adds a summary table, how to use it in TokForge, a file table with sizes from the Hub, and tidier metadata.
README.md
CHANGED
|
@@ -1,68 +1,96 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
language:
|
| 4 |
-
- en
|
| 5 |
tags:
|
| 6 |
-
- tokforge
|
| 7 |
-
-
|
| 8 |
-
-
|
| 9 |
-
-
|
| 10 |
-
-
|
| 11 |
-
-
|
| 12 |
-
-
|
| 13 |
-
-
|
| 14 |
-
-
|
| 15 |
-
pipeline_tag: text-generation
|
| 16 |
inference: false
|
| 17 |
-
base_model: Qwen/Qwen3-0.6B
|
| 18 |
-
base_model_relation: finetune
|
| 19 |
---
|
| 20 |
|
| 21 |
-
# Qwen3
|
|
|
|
|
|
|
| 22 |
|
| 23 |
Published by [TokForge](https://tokforge.ai) for on-device inference.
|
| 24 |
|
| 25 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
-
|
| 28 |
|
| 29 |
-
|
|
|
|
|
|
|
|
|
|
| 30 |
|
| 31 |
-
|
| 32 |
|
| 33 |
-
|
| 34 |
-
- KL-distilled from a `Qwen3-8B` teacher
|
| 35 |
-
- exported as a ready-to-use `MNN` bundle
|
| 36 |
-
- tuned for the classic `CPU draft + GPU/CPU target` TokForge flow
|
| 37 |
|
| 38 |
-
|
|
|
|
|
|
|
| 39 |
|
| 40 |
-
|
| 41 |
-
- Draft threads: `2`
|
| 42 |
-
- Draft predict length: `d=3`
|
| 43 |
-
- Typical target pairing: `Qwen3-8B` in TokForge
|
| 44 |
|
| 45 |
-
##
|
| 46 |
|
| 47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
-
|
| 50 |
|
| 51 |
-
|
|
|
|
| 52 |
|
| 53 |
-
|
|
|
|
|
|
|
|
|
|
| 54 |
|
| 55 |
-
|
| 56 |
-
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
-
|
| 60 |
-
|
| 61 |
-
- ONNX export artifact for reference
|
| 62 |
|
| 63 |
-
##
|
| 64 |
|
| 65 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
|
| 67 |
Typical TokForge recipe:
|
| 68 |
|
|
@@ -91,33 +119,24 @@ Known-good draft-side config:
|
|
| 91 |
}
|
| 92 |
```
|
| 93 |
|
| 94 |
-
##
|
|
|
|
|
|
|
|
|
|
| 95 |
|
| 96 |
-
|
| 97 |
-
- It is a runtime bundle for **TokForge / MNN**.
|
| 98 |
-
- If you want the safest current mobile draft option, this is one of the best places to start.
|
| 99 |
|
| 100 |
-
## Limitations
|
| 101 |
|
| 102 |
-
-
|
| 103 |
-
-
|
| 104 |
- Mobile results depend heavily on backend routing, prompt length, and device thermals.
|
|
|
|
| 105 |
- This is a specialized runtime artifact, not a general-purpose pretrained release.
|
| 106 |
|
| 107 |
-
##
|
| 108 |
-
|
| 109 |
-
- [TokForge Mobile Draft Models](https://huggingface.co/collections/darkmaniac7/tokforge-mobile-draft-models-69c36153ea7084ce78329665)
|
| 110 |
-
|
| 111 |
-
## TokForge
|
| 112 |
-
|
| 113 |
-
- **Website:** https://tokforge.ai
|
| 114 |
-
- **Discord:** https://discord.gg/Acv3CBtfVm
|
| 115 |
-
- **Google Play:** https://play.google.com/store/apps/details?id=dev.tokforge
|
| 116 |
-
- **iOS TestFlight:** https://testflight.apple.com/join/jnufjzRr
|
| 117 |
-
|
| 118 |
-
Runs on-device in the TokForge app.
|
| 119 |
|
| 120 |
-
|
| 121 |
|
| 122 |
## Credits
|
| 123 |
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
base_model: Qwen/Qwen3-0.6B
|
| 4 |
+
base_model_relation: finetune
|
| 5 |
+
pipeline_tag: text-generation
|
| 6 |
+
library_name: mnn
|
| 7 |
language:
|
| 8 |
+
- en
|
| 9 |
tags:
|
| 10 |
+
- tokforge
|
| 11 |
+
- on-device
|
| 12 |
+
- mnn
|
| 13 |
+
- android
|
| 14 |
+
- mobile
|
| 15 |
+
- speculative-decoding
|
| 16 |
+
- qwen3
|
| 17 |
+
- draft-model
|
| 18 |
+
- text-generation
|
|
|
|
| 19 |
inference: false
|
|
|
|
|
|
|
| 20 |
---
|
| 21 |
|
| 22 |
+
# Qwen3 0.6B KL baseline 20k draft (MNN)
|
| 23 |
+
|
| 24 |
+
A research draft: Qwen3 0.6B KL-distilled from Qwen3 8B on 20K teacher samples, as an MNN bundle for speculative decoding experiments.
|
| 25 |
|
| 26 |
Published by [TokForge](https://tokforge.ai) for on-device inference.
|
| 27 |
|
| 28 |
+
| | |
|
| 29 |
+
|---|---|
|
| 30 |
+
| **What it is** | Research draft (draft model for speculative decoding) |
|
| 31 |
+
| **Format** | MNN bundle, 4-bit weights (block 128) |
|
| 32 |
+
| **Runs on** | Android phones (CPU), MNN runtime; not offered in the TokForge app |
|
| 33 |
+
| **Phone class** | No catalog class; needs more free RAM than its 336 MB of weights |
|
| 34 |
+
| **Download** | 340 MB (9 files) |
|
| 35 |
+
| **Base model** | [Qwen/Qwen3-0.6B](https://huggingface.co/Qwen/Qwen3-0.6B) |
|
| 36 |
+
| **License** | [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0) |
|
| 37 |
|
| 38 |
+
## TokForge
|
| 39 |
|
| 40 |
+
- **Website:** https://tokforge.ai
|
| 41 |
+
- **Discord:** https://discord.gg/Acv3CBtfVm
|
| 42 |
+
- **Google Play:** https://play.google.com/store/apps/details?id=dev.tokforge
|
| 43 |
+
- **iOS TestFlight:** https://testflight.apple.com/join/jnufjzRr
|
| 44 |
|
| 45 |
+
Runs on-device in the TokForge app.
|
| 46 |
|
| 47 |
+
## Use it in TokForge
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
+
1. Install TokForge from [Google Play](https://play.google.com/store/apps/details?id=dev.tokforge).
|
| 50 |
+
2. TokForge does not offer or download this research draft. For speculative decoding with Qwen3 chat models, open **Settings > Speed & engine > Models & Modules** and download **Acceleration Pack: Qwen3-0.6B 40k** ([TokForge-AccelerationPack-Draft](https://huggingface.co/darkmaniac7/TokForge-AccelerationPack-Draft)).
|
| 51 |
+
3. To try this draft yourself, run it with MNN's speculative decoding settings below.
|
| 52 |
|
| 53 |
+
Collections: [TokForge Research Drafts](https://huggingface.co/collections/darkmaniac7/tokforge-research-drafts-6af3fde6e8cfcf8046db20b6).
|
|
|
|
|
|
|
|
|
|
| 54 |
|
| 55 |
+
## Files
|
| 56 |
|
| 57 |
+
| File | Size | What it is |
|
| 58 |
+
|---|---:|---|
|
| 59 |
+
| `config.json` | 210 B | MNN runtime settings |
|
| 60 |
+
| `config_cpu.json` | 167 B | Draft runtime settings (CPU, 2 threads, greedy) |
|
| 61 |
+
| `export_args.json` | 1.11 KB | Export settings used for the conversion |
|
| 62 |
+
| `llm.mnn` | 504 KB | Model graph |
|
| 63 |
+
| `llm.mnn.json` | 1.01 MB | Model graph as JSON (reference copy) |
|
| 64 |
+
| `llm.mnn.weight` | 336 MB | Model weights |
|
| 65 |
+
| `llm_config.json` | 4.66 KB | Model settings and chat template |
|
| 66 |
+
| `onnx/llm.onnx` | 933 KB | ONNX graph from the export (reference) |
|
| 67 |
+
| `tokenizer.txt` | 1.61 MB | Tokenizer |
|
| 68 |
|
| 69 |
+
## About the model
|
| 70 |
|
| 71 |
+
Small `Qwen3-0.6B` draft model exported for **TokForge + MNN** speculative decoding on Android. It is meant to be
|
| 72 |
+
paired with larger Qwen3 targets; it is not a chat model on its own.
|
| 73 |
|
| 74 |
+
- trained on `20K` teacher samples
|
| 75 |
+
- KL-distilled from a `Qwen3-8B` teacher
|
| 76 |
+
- exported as a ready-to-use `MNN` bundle
|
| 77 |
+
- tuned for the classic `CPU draft + GPU/CPU target` TokForge flow
|
| 78 |
|
| 79 |
+
| | |
|
| 80 |
+
|---|---|
|
| 81 |
+
| **Architecture** | Qwen3 (dense, full attention) |
|
| 82 |
+
| **Parameters** | 0.6B (4-bit quantized) |
|
| 83 |
+
| **Quantization** | W4A16 (4-bit weights, block size 128) |
|
| 84 |
+
| **Teacher** | [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B) |
|
|
|
|
| 85 |
|
| 86 |
+
## Recommended settings
|
| 87 |
|
| 88 |
+
| Setting | Value |
|
| 89 |
+
|---------|-------|
|
| 90 |
+
| Draft model backend | CPU |
|
| 91 |
+
| Draft threads | 2 |
|
| 92 |
+
| Draft predict length | d=3 |
|
| 93 |
+
| Typical target pairing | Qwen3-8B |
|
| 94 |
|
| 95 |
Typical TokForge recipe:
|
| 96 |
|
|
|
|
| 119 |
}
|
| 120 |
```
|
| 121 |
|
| 122 |
+
## How it was made
|
| 123 |
+
|
| 124 |
+
KL distillation of `Qwen3-0.6B` toward a `Qwen3-8B` teacher on `20K` teacher samples, then exported with MNN's
|
| 125 |
+
`llmexport` with 4-bit weights in blocks of 128 (`export_args.json` has the full settings).
|
| 126 |
|
| 127 |
+
Training acceptance (`alpha`) at the final logged epoch: `0.7178`.
|
|
|
|
|
|
|
| 128 |
|
| 129 |
+
## Limitations
|
| 130 |
|
| 131 |
+
- Research draft: TokForge does not offer it, and we do not publish per-device numbers for it. Results vary by device, target model, and workload.
|
| 132 |
+
- Intended for speculative decoding with larger `Qwen3` targets.
|
| 133 |
- Mobile results depend heavily on backend routing, prompt length, and device thermals.
|
| 134 |
+
- This is **not** a drop-in Transformers repo for HF Inference; it is a runtime bundle for **TokForge / MNN**.
|
| 135 |
- This is a specialized runtime artifact, not a general-purpose pretrained release.
|
| 136 |
|
| 137 |
+
## License
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
|
| 139 |
+
Apache License 2.0 ([text](https://www.apache.org/licenses/LICENSE-2.0)), the license of Qwen3 0.6B and of the Qwen3 8B teacher.
|
| 140 |
|
| 141 |
## Credits
|
| 142 |
|