How to use from the
Use from the
Transformers library
# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Jaidchen/Focus")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)
# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Jaidchen/Focus")
model = AutoModelForCausalLM.from_pretrained("Jaidchen/Focus", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))
Quick Links

Focus

repackaged Qwen 3.6 27B with a simplified architecture and minor opinionated improvements

  • all vision-related components removed
  • reduced storage and memory, faster inference
  • zero loss of output quality

comparison

Qwen 3.6 27B Focus
author Alibaba Qwen Jaid
repository Qwen/Qwen3.6-27B Jaidchen/Focus
architecture qwen3_5 qwen3_5_text
Transformers handler Qwen3_5ForConditionalGeneration Qwen3_5ForCausalLM
tensor entries 1199 866
tensor type bf16 bf16
parameters 27โ€ฏ781โ€ฏ427โ€ฏ952 27โ€ฏ320โ€ฏ697โ€ฏ856
active 100% 100%
vocabulary size 248โ€ฏ320 248โ€ฏ320
context size 262โ€ฏ144 262โ€ฏ144
MTP integrated integrated
sampling strategy random sampling greedy/deterministic
sampling parameters
do_sample: true
temperature: 0.6
top_k: 20
top_p: 0.95
do_sample: false
temperature: 0
top_k: 1
top_p: 1
input modality text, image, video text
repository size 55โ€ฏ586โ€ฏ107โ€ฏ940 54โ€ฏ659โ€ฏ211โ€ฏ447
model size 55โ€ฏ562โ€ฏ855โ€ฏ904 54โ€ฏ641โ€ฏ395โ€ฏ712
splits 15 none
Jinja template Qwen original Qwen original
+ Unsloth tweaks
+ Froggeric tweaks
+ unimodality patch
+ further custom tweaks

pros

  • reduced storage needs
  • reduced loading time
  • reduced VRAM occupancy, thus more room for context
  • increased inference speed
  • simplified architecture, unlocking some further potential for optimizing low-level procedures

cons

  • legally blind
    • Pictures and video frames can still be present in the context without crashing, but their contents are no longer interpreted by the model and wonโ€™t do anything else than waste space.
    • If you occasionally rely on those capabilities, I suggest adding a consult tool to your harness that calls a vision-enabled subagent model like Gemini Flash or GPT.
  • reduced compatibility
    • The simplified architecture is handled by the Qwen3_5ForCausalLM class which may not be included in your inference engine. In this case you would need to ask your agent or integrate it yourself.
      • Inference engines confirmed to have native support:
    • The applied coercions may confuse your inference engine in case it has fixed expectations about the modelโ€™s architecture and thus lead to unpredictable behavior.

caveats

  • model file not split, possibly causing issues if intended to be stored on an HDD from the previous century
  • random sampling disabled by default, less suitable for long-form writing, entertainment and casual chat

Jinja template changes

chat.jinja is reproducibly built from an untouched upstream template plus the ordered patch stack in jinja_build/. jinja_build/build.ps1 applies the patches lexicographically and overwrites the final chat.jinja.

license

Apache 2.0 โ€“ inherited from Qwen 3.6 27B

Downloads last month
196
Safetensors
Model size
27B params
Tensor type
BF16
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for Jaidchen/Focus

Base model

Qwen/Qwen3.6-27B
Finetuned
(440)
this model