Update README.md
Browse files
README.md
CHANGED
|
@@ -1,9 +1,199 @@
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
- open-r1/codeforces-cots
|
| 5 |
-
- nvidia/OpenCodeReasoning
|
| 6 |
-
- patrickfleith/instruction-freak-reasoning
|
| 7 |
base_model:
|
| 8 |
-
- Qwen/Qwen3-0.6B
|
| 9 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: other
|
| 3 |
+
library_name: transformers
|
|
|
|
|
|
|
|
|
|
| 4 |
base_model:
|
| 5 |
+
- Qwen/Qwen3-0.6B
|
| 6 |
+
tags:
|
| 7 |
+
- qwen3
|
| 8 |
+
- code
|
| 9 |
+
- coder
|
| 10 |
+
- reasoning
|
| 11 |
+
- transformers
|
| 12 |
+
- safetensors
|
| 13 |
+
- withinusai
|
| 14 |
+
language:
|
| 15 |
+
- en
|
| 16 |
+
datasets:
|
| 17 |
+
- microsoft/rStar-Coder
|
| 18 |
+
- open-r1/codeforces-cots
|
| 19 |
+
- nvidia/OpenCodeReasoning
|
| 20 |
+
- patrickfleith/instruction-freak-reasoning
|
| 21 |
+
pipeline_tag: text-generation
|
| 22 |
+
---
|
| 23 |
+
|
| 24 |
+
# Qwen3-0.6B-Qrazy-Qoder
|
| 25 |
+
|
| 26 |
+
**Qwen3-0.6B-Qrazy-Qoder** is a compact coding- and reasoning-oriented language model release from **WithIn Us AI**, built on top of **`Qwen/Qwen3-0.6B`** and packaged as a standard **Transformers** checkpoint in **Safetensors** format.
|
| 27 |
+
|
| 28 |
+
This model is intended for lightweight coding assistance, reasoning-style prompt workflows, and compact local or hosted inference where a small model footprint is important.
|
| 29 |
+
|
| 30 |
+
## Model Summary
|
| 31 |
+
|
| 32 |
+
This model is designed for:
|
| 33 |
+
|
| 34 |
+
- code generation
|
| 35 |
+
- code explanation
|
| 36 |
+
- debugging assistance
|
| 37 |
+
- reasoning-oriented coding prompts
|
| 38 |
+
- implementation planning
|
| 39 |
+
- compact instruction following
|
| 40 |
+
- lightweight developer assistant workflows
|
| 41 |
+
|
| 42 |
+
Because this is a **0.6B-class** model, it is best suited for fast, smaller-scope tasks rather than deep long-context reasoning or large multi-file engineering work.
|
| 43 |
+
|
| 44 |
+
## Base Model
|
| 45 |
+
|
| 46 |
+
This model is based on:
|
| 47 |
+
|
| 48 |
+
- **`Qwen/Qwen3-0.6B`**
|
| 49 |
+
|
| 50 |
+
## Training Data / Dataset Lineage
|
| 51 |
+
|
| 52 |
+
The current repository README metadata lists the following datasets:
|
| 53 |
+
|
| 54 |
+
- **`microsoft/rStar-Coder`**
|
| 55 |
+
- **`open-r1/codeforces-cots`**
|
| 56 |
+
- **`nvidia/OpenCodeReasoning`**
|
| 57 |
+
- **`patrickfleith/instruction-freak-reasoning`**
|
| 58 |
+
|
| 59 |
+
These datasets suggest a blend of:
|
| 60 |
+
|
| 61 |
+
- code-focused supervision
|
| 62 |
+
- competitive-programming-style reasoning
|
| 63 |
+
- reasoning-oriented coding data
|
| 64 |
+
- instruction-style reasoning prompts
|
| 65 |
+
|
| 66 |
+
## Intended Use
|
| 67 |
+
|
| 68 |
+
Recommended use cases include:
|
| 69 |
+
|
| 70 |
+
- compact coding assistant experiments
|
| 71 |
+
- short code generation tasks
|
| 72 |
+
- debugging suggestions
|
| 73 |
+
- developer Q&A
|
| 74 |
+
- reasoning-style technical prompting
|
| 75 |
+
- local inference on limited hardware
|
| 76 |
+
- lightweight software workflow support
|
| 77 |
+
|
| 78 |
+
## Suggested Use Cases
|
| 79 |
+
|
| 80 |
+
This model can be useful for:
|
| 81 |
+
|
| 82 |
+
- generating short utility functions
|
| 83 |
+
- explaining code snippets
|
| 84 |
+
- proposing fixes for common bugs
|
| 85 |
+
- creating small implementation plans
|
| 86 |
+
- answering structured coding questions
|
| 87 |
+
- drafting concise technical responses
|
| 88 |
+
|
| 89 |
+
## Out-of-Scope Use
|
| 90 |
+
|
| 91 |
+
This model should not be relied on for:
|
| 92 |
+
|
| 93 |
+
- legal advice
|
| 94 |
+
- medical advice
|
| 95 |
+
- financial advice
|
| 96 |
+
- safety-critical automation
|
| 97 |
+
- autonomous production engineering without review
|
| 98 |
+
- security-critical code without expert validation
|
| 99 |
+
|
| 100 |
+
All generated code should be reviewed, tested, and validated before use.
|
| 101 |
+
|
| 102 |
+
## Repository Contents
|
| 103 |
+
|
| 104 |
+
The repository currently includes standard Hugging Face model assets such as:
|
| 105 |
+
|
| 106 |
+
- `README.md`
|
| 107 |
+
- `.gitattributes`
|
| 108 |
+
- `added_tokens.json`
|
| 109 |
+
- `config.json`
|
| 110 |
+
- `mergekit_config.yml`
|
| 111 |
+
- `merges.txt`
|
| 112 |
+
- `model.safetensors`
|
| 113 |
+
- `special_tokens_map.json`
|
| 114 |
+
- `tokenizer.json`
|
| 115 |
+
- `tokenizer_config.json`
|
| 116 |
+
|
| 117 |
+
## Prompting Guidance
|
| 118 |
+
|
| 119 |
+
This model generally works best when prompts are:
|
| 120 |
+
|
| 121 |
+
- direct
|
| 122 |
+
- scoped to one task
|
| 123 |
+
- explicit about the language or framework
|
| 124 |
+
- clear about whether code, explanation, or both are wanted
|
| 125 |
+
- structured when reasoning is needed
|
| 126 |
+
|
| 127 |
+
### Example prompt styles
|
| 128 |
+
|
| 129 |
+
**Code generation**
|
| 130 |
+
> Write a Python function that removes duplicate records from a JSON list using the `id` field.
|
| 131 |
+
|
| 132 |
+
**Debugging**
|
| 133 |
+
> Explain why this JavaScript function returns `undefined` and provide a corrected version.
|
| 134 |
+
|
| 135 |
+
**Reasoning-oriented coding**
|
| 136 |
+
> Compare two approaches for caching API responses in Python and recommend one.
|
| 137 |
+
|
| 138 |
+
**Implementation planning**
|
| 139 |
+
> Create a step-by-step plan for building a small Flask API with authentication and tests.
|
| 140 |
+
|
| 141 |
+
## Strengths
|
| 142 |
+
|
| 143 |
+
This model may be especially useful for:
|
| 144 |
+
|
| 145 |
+
- compact coding workflows
|
| 146 |
+
- lightweight reasoning prompts
|
| 147 |
+
- low-resource deployments
|
| 148 |
+
- quick iteration
|
| 149 |
+
- structured developer assistance
|
| 150 |
+
- small local inference setups
|
| 151 |
+
|
| 152 |
+
## Limitations
|
| 153 |
+
|
| 154 |
+
Like other compact language models, this model may:
|
| 155 |
+
|
| 156 |
+
- hallucinate APIs or library behavior
|
| 157 |
+
- generate incomplete or incorrect code
|
| 158 |
+
- struggle with long-context tasks
|
| 159 |
+
- make reasoning mistakes on harder prompts
|
| 160 |
+
- require prompt iteration for best results
|
| 161 |
+
- underperform larger coding models on advanced engineering tasks
|
| 162 |
+
|
| 163 |
+
Human review is strongly recommended.
|
| 164 |
+
|
| 165 |
+
## Attribution
|
| 166 |
+
|
| 167 |
+
**WithIn Us AI** is the publisher of this model release.
|
| 168 |
+
|
| 169 |
+
Credit for upstream assets remains with their original creators, including:
|
| 170 |
+
|
| 171 |
+
- **Qwen** for **`Qwen/Qwen3-0.6B`**
|
| 172 |
+
- **Microsoft** for **`microsoft/rStar-Coder`**
|
| 173 |
+
- the creators of **`open-r1/codeforces-cots`**
|
| 174 |
+
- **NVIDIA** for **`nvidia/OpenCodeReasoning`**
|
| 175 |
+
- **patrickfleith** for **`patrickfleith/instruction-freak-reasoning`**
|
| 176 |
+
|
| 177 |
+
## License
|
| 178 |
+
|
| 179 |
+
This draft uses:
|
| 180 |
+
|
| 181 |
+
- `license: other`
|
| 182 |
+
|
| 183 |
+
If you maintain this repo, replace this with the exact license terms you want displayed and ensure they align with any upstream licensing requirements.
|
| 184 |
+
|
| 185 |
+
## Acknowledgments
|
| 186 |
+
|
| 187 |
+
Thanks to:
|
| 188 |
+
|
| 189 |
+
- **WithIn Us AI**
|
| 190 |
+
- **Qwen**
|
| 191 |
+
- **Microsoft**
|
| 192 |
+
- **NVIDIA**
|
| 193 |
+
- the dataset creators listed above
|
| 194 |
+
- the Hugging Face ecosystem
|
| 195 |
+
- the broader open-source AI community
|
| 196 |
+
|
| 197 |
+
## Disclaimer
|
| 198 |
+
|
| 199 |
+
This model may produce inaccurate, insecure, incomplete, or misleading outputs. All important generations, especially code and technical guidance, should be reviewed and tested before real-world use.
|