Instructions to use AbstractPhil/mini-beatrix-1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AbstractPhil/mini-beatrix-1 with Transformers:

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

pipe = pipeline("text-generation", model="AbstractPhil/mini-beatrix-1", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("AbstractPhil/mini-beatrix-1", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AbstractPhil/mini-beatrix-1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AbstractPhil/mini-beatrix-1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AbstractPhil/mini-beatrix-1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/AbstractPhil/mini-beatrix-1

SGLang

How to use AbstractPhil/mini-beatrix-1 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 "AbstractPhil/mini-beatrix-1" \
    --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": "AbstractPhil/mini-beatrix-1",
		"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 "AbstractPhil/mini-beatrix-1" \
        --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": "AbstractPhil/mini-beatrix-1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use AbstractPhil/mini-beatrix-1 with Docker Model Runner:
```
docker model run hf.co/AbstractPhil/mini-beatrix-1
```

AbstractPhil commited on Aug 26

Commit

2ea0f30

verified ·

1 Parent(s): 34d7ec4

mirror 0.7.5: 2s mission pivot (AlephLLM@73218bd)

Browse files

Files changed (1) hide show

presets.py +8 -3

presets.py CHANGED Viewed

@@ -174,8 +174,13 @@ PRESETS: dict[str, Preset] = {
                           governor="minsep", governor_theta=45.0),
         curriculum=_curriculum(500_000_000, 8_000_000_000, 16_000_000_000),
     ),
-    # The lawful screen craft for the same era: every gating cell (full-splat
-    # causal viability, constellation composition, D ladder) runs here first.
     "mini-beatrix-2s": Preset(
         model=AlephLMConfig(name="mini-beatrix-2s", d_model=1024, n_layers=20,
                             n_heads=16, context=4096,
@@ -183,7 +188,7 @@ PRESETS: dict[str, Preset] = {
                             hub_K=64, hub_D=128, hub_const=4,
                             bank_experts=3, bank_ff=1024,
                             head_K=256, head_D=256, hub_chunk=256,
-                            hub_ckpt=2),
         train=TrainConfig(micro_batch=16, grad_accum=4,
                           governor="minsep", governor_theta=45.0),
         curriculum=_curriculum(300_000_000, 5_000_000_000, 10_000_000_000),

                           governor="minsep", governor_theta=45.0),
         curriculum=_curriculum(500_000_000, 8_000_000_000, 16_000_000_000),
     ),
+    # THE ACTIVE MISSION (2026-08-26, Phil: "train the next stage up from
+    # the beatrix v1; we can't train the large one currently"): the lawful
+    # full-splat craft one rung above v1 — d1024 L20 ctx4096, governed
+    # 4x64@128 books (4x supply headroom vs v1's crowded 16x). Also the
+    # screen bed for every v2-era gating cell. hub_ckpt=0: at 237M the
+    # retained scan fits the 96GB card, so the recompute tax is pure waste
+    # (fallback: set hub_ckpt=2 if the preflight bench gate aborts >88GB).
     "mini-beatrix-2s": Preset(
         model=AlephLMConfig(name="mini-beatrix-2s", d_model=1024, n_layers=20,
                             n_heads=16, context=4096,
                             hub_K=64, hub_D=128, hub_const=4,
                             bank_experts=3, bank_ff=1024,
                             head_K=256, head_D=256, hub_chunk=256,
+                            hub_ckpt=0),
         train=TrainConfig(micro_batch=16, grad_accum=4,
                           governor="minsep", governor_theta=45.0),
         curriculum=_curriculum(300_000_000, 5_000_000_000, 10_000_000_000),