Qwen2.5 7B, self-quantized to GGUF by Atomic Chat. Built straight from Qwen's original weights with a per-tensor importance matrix, so this is not a repack of somebody else's files. Runs fully offline.
Highlights
- 7.6B parameters: the weights this repo quantizes.
- Context length: 32,768 tokens (32K), as published by Qwen.
- 28 layers: Dense decoder, hybrid sliding-window (131072) and global attention.
- Full imatrix ladder: every quant is calibrated with an importance matrix.
- Multilingual support: for over 29 languages, including Chinese, English, French, Spanish, Portuguese, German, Italian, Russian, Japanese, Korean, Vietnamese, Thai, Arabic, and more.
[!NOTE] These GGUFs are self-quantized from the original weights, not a repack. The importance matrix keeps low-bit quants closer to the full-precision model.
[!IMPORTANT] Always pass
--jinjaso the Qwen2.5 7B chat template is applied. Without it the model can emit malformed turns.
Model Overview
| Property | Value |
|---|---|
| Base model | Qwen/Qwen2.5-7B-Instruct |
| Parameters | 7.6B |
| Layers | 28 |
| Sliding window | 131072 tokens |
| Context length | 32,768 tokens (32K) |
| Vocabulary | 152,064 |
| Modalities | Text |
| Architecture | Dense decoder, hybrid sliding-window (131072) and global attention, 28 attention heads over 4 KV heads, Qwen2ForCausalLM |
| This repo | GGUF quants (imatrix). Quants: Q4_K_M, UD-Q4_K_XL, Q5_K_M, Q6_K, Q8_0 |
Choosing a quant
| Quant | Size | Notes |
|---|---|---|
Q4_K_M | 4.7 GB | Recommended default. Best balance of size, speed and quality. |
UD-Q4_K_XL | 5.1 GB | Dynamic. Embeddings and output kept at Q8_0 for higher quality at a Q4 footprint. |
Q5_K_M | 5.4 GB | Higher quality, low loss. |
Q6_K | 6.3 GB | Near lossless, noticeably lighter than Q8_0. |
Q8_0 | 8.1 GB | Effectively lossless, reference quality. |
[!TIP] Pick the largest file that fits your (V)RAM with room for context.
Q4_K_MorUD-Q4_K_XLis the sweet spot for most setups;Q6_KorQ8_0for maximum fidelity.
Get started
Run Qwen2.5 7B locally with:
- Atomic Chat: the easiest path. Open the app, search
AtomicChat/Qwen2.5-7B-Instruct-GGUF, pick a quant, hit Use this model. - llama.cpp:
llama-server -hf AtomicChat/Qwen2.5-7B-Instruct-GGUF:Q4_K_M --jinja -c 8192 - Ollama:
ollama run hf.co/AtomicChat/Qwen2.5-7B-Instruct-GGUF:Q4_K_M - LM Studio / Jan: search the repo id, download any quant.
Best practices
| Parameter | Value |
|---|---|
| temperature | 0.7 |
| top_p | 0.8 |
| top_k | 20 |
| repetition_penalty | 1.05 |
Qwen's recommended sampling configuration for Qwen/Qwen2.5-7B-Instruct.
Run in llama.cpp
git clone https://github.com/ggml-org/llama.cpp
cmake llama.cpp -B llama.cpp/build -DBUILD_SHARED_LIBS=OFF -DGGML_CUDA=ON
cmake --build llama.cpp/build --config Release -j --target llama-cli llama-server
./llama.cpp/build/bin/llama-server \
-hf AtomicChat/Qwen2.5-7B-Instruct-GGUF:Q4_K_M \
--jinja -ngl 99 -c 8192 -fa on
How these were made
- Download
Qwen/Qwen2.5-7B-Instruct(original weights). - Convert to f16 GGUF with llama.cpp.
- Build an importance matrix over our calibration corpus.
- Quantize the ladder with
--imatrix. UD-Q4_K_XLadditionally pins the token-embedding and output tensors toQ8_0.
License
Original model by Qwen, released under the Apache 2.0 license. Full terms: Apache 2.0. Quantized by Atomic Chat.