mradermacher

/

model_requests

Like 224

can this model please be reuploaded to huggingface?

已经在huggingface上了 https://huggingface.co/hutaobentao/furina-qwen3.6-27b-merged-2.0

I'll see if I can find the time. What kind of quantification do you need?

Greetings,
Sebastian

如果能全量化最好了,谢谢

Hey, I tried to convert your model to GGUF for you, but I ran into a small issue.

It looks like the tokenizer.json and tokenizer_config.json files are missing in your repository. Without the tokenizer files, the conversion script (llama.cpp) can't package the vocabulary, and the model won't know how to process text.

Since your model is based on Qwen3.6-27B, I'm currently using the tokenizer files from the base model as a workaround to finish the GGUF conversion.

If you have a moment, it would be great if you could upload the tokenizer files to your repository! This will make it much easier for everyone to use and convert your model in the future.

Hey there! Just a quick update: I'm currently converting and uploading all GGUF variants (Q4_K_M, Q6_K, Q8_0, and F16) to a community repository here:
👉 Earlychildhoodeducation/furina-qwen3.6-27b-merged-2.0-GGUF

Please be a bit patient until later this evening, as my upload bandwidth is currently quite busy. Once it's up, feel free to give it a test and let me know if everything works smoothly for you!

Warm regards to China from Schifferstadt ;)

Hi @Mightysweet , if you have time you can upload the fixed safetensors repo so that I can queue it for team mradermacher =)

i thought you try to avoid double work (unless ur doing i1 or other quants mighty didnt upload)

Hi everyone! Quick update: Everything is fully uploaded now.

To address the point about double work: It wasn't actually redundant, because the original repository was missing all the essential tokenizer files and config parts, causing automated conversion scripts (like llama.cpp) to fail out of the box.

I've now uploaded the complete, fixed Safetensors repository plus the missing tokenizer files (sourced from the base model) alongside my locally converted GGUF variants (Q4, Q6, Q8, F16) all in one place:
👉 Earlychildhoodeducation/furina-qwen3.6-27b-merged-2.0-GGUF

So
@RichardErkhov , the repo is fully fixed and ready for the team mradermacher imatrix queue now.
And @NILKNARFGonzo , thanks for keeping an eye on efficiency—fixing the broken tokenizer first was the necessary step to make any automated or manual quantization possible in the first place!
and @XingChina , You can now use the gguf files directly. The F16 will be online in about an hour.

Cheers from Schifferstadt!

And @NILKNARFGonzo , thanks for keeping an eye on efficiency—fixing the broken tokenizer first was the necessary step to make any automated or manual quantization possible in the first place!

unless ur quanting my models that python script update was unnecessary

非常感谢你们🙏

Im not sure what the url will be, but its queued lol

You can check for progress at http://hf.tst.eu/status.html or regularly check the model
summary page at https://hf.tst.eu/model#furina-qwen3.6-27b-merged-2.0-GGUF-GGUF or maybe https://hf.tst.eu/model#furina-qwen3.6-27b-merged-2.0-GGUF for quants to appear.

let me know if it works since I am not very active right now and will for sure forget