Instructions to use SvalTek/L3-CharThink-Base-Fix with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SvalTek/L3-CharThink-Base-Fix with Transformers:

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

pipe = pipeline("text-generation", model="SvalTek/L3-CharThink-Base-Fix")
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("SvalTek/L3-CharThink-Base-Fix")
model = AutoModelForCausalLM.from_pretrained("SvalTek/L3-CharThink-Base-Fix", 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 SvalTek/L3-CharThink-Base-Fix with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SvalTek/L3-CharThink-Base-Fix"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SvalTek/L3-CharThink-Base-Fix",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SvalTek/L3-CharThink-Base-Fix

SGLang

How to use SvalTek/L3-CharThink-Base-Fix 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 "SvalTek/L3-CharThink-Base-Fix" \
    --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": "SvalTek/L3-CharThink-Base-Fix",
		"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 "SvalTek/L3-CharThink-Base-Fix" \
        --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": "SvalTek/L3-CharThink-Base-Fix",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Desktop
Docker Model Runner
How to use SvalTek/L3-CharThink-Base-Fix with Docker Model Runner:
```
docker model run hf.co/SvalTek/L3-CharThink-Base-Fix
```

Theros commited on May 26

Commit

ca8bdbe

verified ·

1 Parent(s): ce4fe2e

Update README.md

Browse files

Files changed (1) hide show

README.md +8 -2

README.md CHANGED Viewed

@@ -10,10 +10,16 @@ language:
 - en
 ---
-This model is tuned for character-POV internal reasoning.
-Its `<think>` blocks are intended to represent subjective in-character cognition, memory, emotional appraisal, and scene-state processing, not generic assistant planning.
 ---

 - en
 ---
+This model is tuned for character-POV internal cognition.
+Its `<think>` blocks are intended to represent subjective in-character cognition: perception, memory, emotional appraisal, motive formation, and scene-state processing. They are not generic assistant planning or task-solving chain-of-thought.
+For best results, give the model a system prompt that establishes an identity-bearing frame, such as a named persona, character card, role, narrator voice, or immersive writing mode. Minimal examples include:
+"You write immersive character responses.
+"You are {character name}, {brief identity/personality/context}."
+Without an identity-bearing system prompt, the base assistant/RP prior may dominate and the model may produce ordinary visible roleplay without opening a `<think>` block.
 ---