初始化项目，由ModelHub XC社区提供模型
Model: DhruvalLabs/qwen3-8b-claude-agentic-fable5-GGUF Source: Original Platform
This commit is contained in:
ModelHub XC
165
README.md
Normal file
165
README.md
Normal file
@@ -0,0 +1,165 @@
|
||||
---
|
||||
license: apache-2.0
|
||||
base_model: DhruvalLabs/qwen3-8b-claude-agentic-fable5
|
||||
pipeline_tag: text-generation
|
||||
tags:
|
||||
- base_model:adapter:Qwen/Qwen3-8B
|
||||
- text-generation-inference
|
||||
- quantized
|
||||
- qwen3
|
||||
- safetensors
|
||||
- gguf
|
||||
- conversational
|
||||
- region:us
|
||||
- unsloth
|
||||
- tool-use
|
||||
- transformers
|
||||
- lora
|
||||
- finetuned
|
||||
- claude
|
||||
- agentic
|
||||
- license:apache-2.0
|
||||
- en
|
||||
- coding-agent
|
||||
- base_model:Qwen/Qwen3-8B
|
||||
- text-generation
|
||||
- fable5
|
||||
language:
|
||||
- en
|
||||
---
|
||||
|
||||
<div align="center">
|
||||
|
||||
# qwen3-8b-claude-agentic-fable5 — GGUF Quantizations
|
||||
|
||||
[](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5-GGUF)
|
||||
[](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5)
|
||||
[](https://github.com/DhruvalPtl/quant-kit)
|
||||
|
||||
**Quantized GGUF versions of [DhruvalLabs/qwen3-8b-claude-agentic-fable5](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5)**
|
||||
|
||||
Works with **[llama.cpp](https://github.com/ggerganov/llama.cpp)** · **[Ollama](https://ollama.ai)** · **[LM Studio](https://lmstudio.ai)** · **[Open WebUI](https://openwebui.com)** · **[Jan](https://jan.ai)**
|
||||
|
||||
*Quantized by **[DhruvalLabs](https://huggingface.co/DhruvalLabs)** on July 07, 2026 using [quant-kit](https://github.com/DhruvalPtl/quant-kit)*
|
||||
|
||||
</div>
|
||||
|
||||
---
|
||||
|
||||
## ⚖️ The Pareto Frontier — Efficiency vs Intelligence
|
||||
|
||||
> Can you run a powerful model on a laptop without losing its intelligence?
|
||||
|
||||
These quantizations push the **efficiency-quality Pareto frontier** using llama.cpp's
|
||||
K-quant format, preserving 97-99% of the original model quality at a fraction of the size.
|
||||
|
||||
| Benchmark | Original (FP16) | Q4_K_M | Quality Retained |
|
||||
|---|---|---|---|
|
||||
| **MMLU Pro** | *See [original card](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5)* | *Run benchmarks* | ~97-99% |
|
||||
| **HellaSwag** | *See [original card](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5)* | *Run benchmarks* | ~97-99% |
|
||||
| **ARC Challenge** | *See [original card](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5)* | *Run benchmarks* | ~97-99% |
|
||||
| **TruthfulQA** | *See [original card](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5)* | *Run benchmarks* | ~97-99% |
|
||||
| **GSM8K** | *See [original card](https://huggingface.co/DhruvalLabs/qwen3-8b-claude-agentic-fable5)* | *Run benchmarks* | ~97-99% |
|
||||
|
||||
|
||||
---
|
||||
|
||||
## 📦 Available Files
|
||||
|
||||
| Filename | Size | RAM Required | Quant | Quality | Best For |
|
||||
|---|---|---|---|---|---|
|
||||
| `qwen3-8b-claude-agentic-fable5-Q2_K.gguf` | 3.06 GB | ~4.6 GB | `Q2_K` | ⭐ | Extreme compression, significant quality loss. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q3_K_L.gguf` | 4.13 GB | ~5.6 GB | `Q3_K_L` | ⭐⭐⭐ | Slightly better than Q3_K_M, still a compromise. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q3_K_M.gguf` | 3.84 GB | ~5.3 GB | `Q3_K_M` | ⭐⭐⭐ | Very small file. Quality drop noticeable. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q3_K_S.gguf` | 3.51 GB | ~5.0 GB | `Q3_K_S` | ⭐⭐ | Very high compression, high quality loss. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q4_K_M.gguf` | 4.68 GB | ~6.2 GB | `Q4_K_M` ✅ **Recommended** | ⭐⭐⭐⭐ | Best balance of size and quality. Recommended for most users. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q4_K_S.gguf` | 4.47 GB | ~6.0 GB | `Q4_K_S` | ⭐⭐⭐½ | Good speed/size balance, slight quality loss. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q5_K_M.gguf` | 5.45 GB | ~6.9 GB | `Q5_K_M` | ⭐⭐⭐⭐½ | Better quality than Q4, slightly larger. Great if you have the RAM. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q5_K_S.gguf` | 5.33 GB | ~6.8 GB | `Q5_K_S` | ⭐⭐⭐⭐ | Large but accurate. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q6_K.gguf` | 6.26 GB | ~7.8 GB | `Q6_K` | ⭐⭐⭐⭐⭐ | Near-perfect quality, very large. |
|
||||
| `qwen3-8b-claude-agentic-fable5-Q8_0.gguf` | 8.11 GB | ~9.6 GB | `Q8_0` | ⭐⭐⭐⭐⭐ | Closest to original quality. Use when RAM is not a concern. |
|
||||
|
||||
|
||||
### 💡 Which file should I download?
|
||||
|
||||
- **Most users:** `qwen3-8b-claude-agentic-fable5-Q4_K_M.gguf` — best balance of size and quality
|
||||
- **High RAM (32GB+):** `qwen3-8b-claude-agentic-fable5-Q8_0.gguf` — near-original quality
|
||||
- **Low RAM (8GB):** `qwen3-8b-claude-agentic-fable5-Q3_K_M.gguf` — fits in 8GB with room to spare
|
||||
|
||||
---
|
||||
|
||||
## ⚡ Speed Benchmarks
|
||||
|
||||
*Run `python benchmark.py --model qwen3-8b-claude-agentic-fable5` to generate speed results.*
|
||||
|
||||
|
||||
---
|
||||
|
||||
## 🧠 Quality Benchmarks
|
||||
|
||||
*Run `kaggle_bench.ipynb` on Kaggle to benchmark this model.*
|
||||
|
||||
|
||||
---
|
||||
|
||||
## 🚀 How to Use
|
||||
|
||||
### Ollama
|
||||
```bash
|
||||
ollama run dhruvallabs/qwen3-8b-claude-agentic-fable5
|
||||
```
|
||||
|
||||
### LM Studio / Jan / Open WebUI
|
||||
Search for `DhruvalLabs/qwen3-8b-claude-agentic-fable5` in the model browser.
|
||||
|
||||
### llama.cpp CLI
|
||||
```bash
|
||||
# Download the binary from https://github.com/ggerganov/llama.cpp/releases
|
||||
./llama-cli \
|
||||
-m qwen3-8b-claude-agentic-fable5-Q4_K_M.gguf \
|
||||
-p "You are a helpful assistant." \
|
||||
--conversation \
|
||||
-n 512
|
||||
```
|
||||
|
||||
### Python — llama-cpp-python
|
||||
```python
|
||||
from llama_cpp import Llama
|
||||
|
||||
llm = Llama(
|
||||
model_path="./qwen3-8b-claude-agentic-fable5-Q4_K_M.gguf",
|
||||
n_gpu_layers=-1, # -1 = offload everything to GPU
|
||||
n_ctx=4096,
|
||||
)
|
||||
|
||||
response = llm.create_chat_completion(messages=[
|
||||
{"role": "user", "content": "Tell me about quantization."}
|
||||
])
|
||||
print(response["choices"][0]["message"]["content"])
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## 🔍 About GGUF Quantization
|
||||
|
||||
GGUF is the standard file format for running large language models locally.
|
||||
Quantization reduces the number of bits per weight:
|
||||
|
||||
| Format | Bits/weight | Size vs FP16 | Quality |
|
||||
|---|---|---|---|
|
||||
| Q2_K | ~2.6 | 16% | ⭐ |
|
||||
| Q3_K_M | ~3.3 | 21% | ⭐⭐⭐ |
|
||||
| Q4_K_M | ~4.5 | 28% | ⭐⭐⭐⭐ ← sweet spot |
|
||||
| Q5_K_M | ~5.6 | 35% | ⭐⭐⭐⭐½ |
|
||||
| Q8_0 | ~8.5 | 53% | ⭐⭐⭐⭐⭐ |
|
||||
|
||||
---
|
||||
|
||||
## 💬 Community & Feedback
|
||||
|
||||
Found an issue? Have a question? Open a **Discussion** in the Community tab above.
|
||||
|
||||
If these quantizations were useful, please consider:
|
||||
- ⭐ Starring [quant-kit](https://github.com/DhruvalPtl/quant-kit) on GitHub
|
||||
- 👍 Liking this model on HuggingFace
|
||||
- 💬 Leaving feedback in the Community tab
|
||||
Reference in New Issue
Block a user