Instructions to use BernaLabs/berna-r8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use BernaLabs/berna-r8 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="BernaLabs/berna-r8")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("BernaLabs/berna-r8", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use BernaLabs/berna-r8 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "BernaLabs/berna-r8" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "BernaLabs/berna-r8", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/BernaLabs/berna-r8
- SGLang
How to use BernaLabs/berna-r8 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 "BernaLabs/berna-r8" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "BernaLabs/berna-r8", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'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 "BernaLabs/berna-r8" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "BernaLabs/berna-r8", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use BernaLabs/berna-r8 with Docker Model Runner:
docker model run hf.co/BernaLabs/berna-r8
Download code/cells/dna_kernel_fix.py from BernaLabs/berna-r8: direct link, hf CLI and curl.
- Browser
- Download file 1.72 kB
-
https://huggingface.co/BernaLabs/berna-r8/resolve/main/code/cells/dna_kernel_fix.py
- Command line
-
hf download hf://BernaLabs/berna-r8/code/cells/dna_kernel_fix.py
-
curl -L -o dna_kernel_fix.py https://huggingface.co/BernaLabs/berna-r8/resolve/main/code/cells/dna_kernel_fix.py
1.72 kB
| from pathlib import Path | |
| p = Path("/data/berna-r7/core/code/cells/dna_kernel.py") | |
| text = p.read_text() | |
| old = '''def decide(dna: DNA, S: float) -> dict: | |
| """ | |
| (D_c, K_c, u_c) -> {route, grow, split, retain} | |
| Returns a dict of signals + a discrete decision. | |
| """ | |
| route = signal_route(dna) | |
| grow = signal_grow(dna) | |
| split = signal_split(dna, S) | |
| retain = signal_retain(dna) | |
| signals = {"route": route, "grow": grow, "split": split, "retain": retain} | |
| # Discrete decision: highest signal wins | |
| decision = max(signals, key=signals.get) | |
| signals["decision"] = decision | |
| return signals''' | |
| new = '''def decide(dna: DNA, S: float, S_star: float = 0.85, | |
| t_consecutive: int = 3) -> dict: | |
| """ | |
| (D_c, K_c, u_c) -> {route, grow, split, retain} | |
| Priority-based decision per R5 Section 3.5. | |
| Priority order (per R5): | |
| 1. split - if S >= S* sustained (handled by caller's S_history) | |
| 2. retain - if persistence high (frozen state) | |
| 3. grow - if plasticity high and not saturated | |
| 4. route - default behavior | |
| """ | |
| route = signal_route(dna) | |
| grow = signal_grow(dna) | |
| split = signal_split(dna, S, S_star) | |
| retain = signal_retain(dna) | |
| signals = {"route": route, "grow": grow, "split": split, "retain": retain} | |
| # Priority-based decision (R5 Section 3.5) | |
| if S >= S_star: | |
| decision = "split" | |
| elif retain > 0.7: | |
| decision = "retain" | |
| elif grow > 0.6: | |
| decision = "grow" | |
| else: | |
| decision = "route" | |
| signals["decision"] = decision | |
| return signals''' | |
| assert old in text, "pattern not found" | |
| text = text.replace(old, new, 1) | |
| p.write_text(text) | |
| print("✅ decide() patched") | |