Instructions to use Jaidchen/Focus-Red with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Jaidchen/Focus-Red with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Jaidchen/Focus-Red")
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-Red")
model = AutoModelForCausalLM.from_pretrained("Jaidchen/Focus-Red", 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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Jaidchen/Focus-Red with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Jaidchen/Focus-Red"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Jaidchen/Focus-Red",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Jaidchen/Focus-Red

SGLang

How to use Jaidchen/Focus-Red with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Jaidchen/Focus-Red" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Jaidchen/Focus-Red",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Jaidchen/Focus-Red" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Jaidchen/Focus-Red",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Jaidchen/Focus-Red with Docker Model Runner:
```
docker model run hf.co/Jaidchen/Focus-Red
```

Jaidchen commited on Aug 15

Commit

495e0b0

verified ·

1 Parent(s): 85f197c

Update README.md

Browse files

Files changed (1) hide show

README.md +33 -8

README.md CHANGED Viewed

@@ -16,7 +16,7 @@ tags:
 <img style='height: 16em' src='https://cdn-uploads.huggingface.co/production/uploads/63e9a1c4ccae1fe5c622b9f6/zOXMkdhMfeQWxDq6JHj0B.webp'/>
 </center>
-# Focus-Red
 repackaged [Qwen 3.8 27B](https://huggingface.co/Qwen/Qwen3.8-27B) with a simplified architecture and minor opinionated improvements
@@ -30,7 +30,7 @@ repackaged [Qwen 3.8 27B](https://huggingface.co/Qwen/Qwen3.8-27B) with a simpli
 <tr>
   <th></th>
   <th>Qwen 3.8 27B</th>
-  <th>Focus-Red</th>
 </tr>
 <tr>
   <td>author</td>
@@ -86,10 +86,10 @@ repackaged [Qwen 3.8 27B](https://huggingface.co/Qwen/Qwen3.8-27B) with a simpli
   <td>262 144</td>
   <td>262 144</td>
 </tr>
-<tr style='opacity: 50%'>
   <td>MTP</td>
   <td>integrated</td>
-  <td>detached → <a href='https://huggingface.co/Jaidchen/Focus-Red-MTP'>Focus-Red-MTP</a></td>
 </tr>
 <tr>
   <td>sampling strategy</td>
@@ -163,10 +163,35 @@ repackaged [Qwen 3.8 27B](https://huggingface.co/Qwen/Qwen3.8-27B) with a simpli
 ## Jinja template
-This build uses `dist/chat_template.jinja` from [jaidlab/focus-chat-template](https://github.com/jaidlab/focus-chat-template). The template is reproducibly built from Qwen/Qwen3.8-27B's pinned upstream template at commit `1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0` plus the repository's ordered patch stack.
-- template SHA-256: `5c381ca45e9538c7a2406331b554ee7d62cf3d0b8c115f17687d4fdd5590a239`
-- template size: 9,710 bytes
 ## license

 <img style='height: 16em' src='https://cdn-uploads.huggingface.co/production/uploads/63e9a1c4ccae1fe5c622b9f6/zOXMkdhMfeQWxDq6JHj0B.webp'/>
 </center>
+# Focus Red
 repackaged [Qwen 3.8 27B](https://huggingface.co/Qwen/Qwen3.8-27B) with a simplified architecture and minor opinionated improvements
 <tr>
   <th></th>
   <th>Qwen 3.8 27B</th>
+  <th>Focus Red</th>
 </tr>
 <tr>
   <td>author</td>
   <td>262 144</td>
   <td>262 144</td>
 </tr>
+<tr>
   <td>MTP</td>
   <td>integrated</td>
+  <td><a href='https://huggingface.co/Jaidchen/Focus-Red-MTP'>Jaidchen/Focus-Red-MTP</a></td>
 </tr>
 <tr>
   <td>sampling strategy</td>
 ## Jinja template
+- base: [Qwen/Qwen3.8-27B original](https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/chat_template.jinja)
+  - Retains Qwen3.8's native `reasoning_effort` handling (`xhigh` by default, plus `medium` and `low`) and its reasoning instructions. ([lines 45–59](https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/chat_template.jinja#L45-L59))
+  - Retains injection of those reasoning instructions into the tool/system preamble whenever thinking is enabled. ([lines 60–94](https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/chat_template.jinja#L60-L94))
+  - Preserves Qwen3.8's default-on historical thinking behavior while the later Froggeric layer avoids synthesizing empty thinking blocks. ([lines 108–120](https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/chat_template.jinja#L108-L120))
+- adopted tweaks from [Unsloth Qwen3.8](https://huggingface.co/unsloth/Qwen3.8-27B/blob/3ea932cee0a432ae86e9c7826cbe8aef52323a28/chat_template.jinja)
+  - commit `3ea932cee0a432ae86e9c7826cbe8aef52323a28`
+  - Added `developer` role as alias for `system` and merged leading system/developer messages into a single policy message. ([lines 45–59](https://huggingface.co/unsloth/Qwen3.8-27B/blob/3ea932cee0a432ae86e9c7826cbe8aef52323a28/chat_template.jinja#L45-L59), [lines 80–93](https://huggingface.co/unsloth/Qwen3.8-27B/blob/3ea932cee0a432ae86e9c7826cbe8aef52323a28/chat_template.jinja#L80-L93))
+  - Allowed histories without a normal human query. ([lines 95–107](https://huggingface.co/unsloth/Qwen3.8-27B/blob/3ea932cee0a432ae86e9c7826cbe8aef52323a28/chat_template.jinja#L95-L107))
+- retained compatibility tweak from the previous [Unsloth Qwen3.6 layer](https://huggingface.co/unsloth/Qwen3.6-27B/blob/d6e694245c1d535508bc54062f5e258d7b1e04f5/chat_template.jinja)
+  - Made mapping-argument rendering portable by avoiding `|items`; the current Unsloth Qwen3.8 template uses `|items` again, so Focus deliberately keeps the portable implementation. ([lines 122–130](https://huggingface.co/unsloth/Qwen3.6-27B/blob/d6e694245c1d535508bc54062f5e258d7b1e04f5/chat_template.jinja#L122-L130))
+- adopted tweaks from [Froggeric Qwen3.8 version](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja)
+  - commit `9f14778c92c3b5ed3e0738085694c0d3452802dd`
+  - Retained `preserve_thinking` semantics and avoided rendering empty historical thinking blocks. ([lines 8–15](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L8-L15), [lines 227–233](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L227-L233))
+  - Retained mid-conversation system/developer messages as Qwen system turns. ([lines 172–183](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L172-L183))
+  - Allowed string-valued `message.thinking` as fallback for historical reasoning. ([lines 186–199](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L186-L199))
+  - Applied boundary-aware `</think>` parsing. ([lines 200–226](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L200-L226))
+  - Supported both wrapped and direct tool calls, but never null-wrapped. ([lines 234–241](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L234-L241))
+  - Preserved non-empty string tool arguments. ([lines 248–280](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L248-L280))
+  - Used direct message indexing for tool-response grouping instead of `loop.previtem`/`loop.nextitem`. ([lines 298–317](https://huggingface.co/froggeric/Qwen-Fixed-Chat-Templates/blob/9f14778c92c3b5ed3e0738085694c0d3452802dd/chat_template.jinja#L298-L317))
+- custom tweaks
+  - Merged any number of leading system/developer messages, generalizing the Unsloth merge behavior.
+  - Kept Qwen3.8's reasoning instructions ahead of both merged policy text and the tools preamble.
+  - Applied boundary-aware `</think>` parsing but excluded Froggeric's malformed-tag recovery.
+  - Raised an error containing the offending role instead of Qwen's generic unknown-role error. ([lines 160–162](https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/chat_template.jinja#L160-L162))
+  - Used direct message indexing for tool-response grouping without Froggeric's error-escalation state.
+  - Guarded undefined `tools`/`tool_calls` and handled non-mapping content items defensively.
+  - Retained Qwen's proposed `continue_final_message` fix for partial assistant prefills. ([lines 130–132](https://huggingface.co/Qwen/Qwen3.6-27B/blob/0774bb84a25294dac0538c2bfc7abc08ae89d522/chat_template.jinja#L130-L132))
+- unimodality patch
+  - Removed Qwen3.8's vision-token machinery and rendered image/video content as `[image]` and `[video]`. ([lines 1–41](https://huggingface.co/Qwen/Qwen3.8-27B/blob/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0/chat_template.jinja#L1-L41))
 ## license