Instructions to use osunlp/PreNormLoop-0.6B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use osunlp/PreNormLoop-0.6B with Transformers:

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

pipe = pipeline("text-generation", model="osunlp/PreNormLoop-0.6B", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("osunlp/PreNormLoop-0.6B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use osunlp/PreNormLoop-0.6B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "osunlp/PreNormLoop-0.6B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "osunlp/PreNormLoop-0.6B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/osunlp/PreNormLoop-0.6B

SGLang

How to use osunlp/PreNormLoop-0.6B 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 "osunlp/PreNormLoop-0.6B" \
    --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": "osunlp/PreNormLoop-0.6B",
		"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 "osunlp/PreNormLoop-0.6B" \
        --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": "osunlp/PreNormLoop-0.6B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use osunlp/PreNormLoop-0.6B with Docker Model Runner:
```
docker model run hf.co/osunlp/PreNormLoop-0.6B
```

ykyao commited on 5 days ago

Commit

614ad44

verified ·

1 Parent(s): 371e22c

Simplify model card and hide internal checkpoint filenames

Browse files

Files changed (1) hide show

README.md +8 -55

README.md CHANGED Viewed

@@ -14,19 +14,13 @@ tags:
 # ARLoop-baseline-0.6B
 This is the fixed-loop PreNorm baseline for ARLoop. This release
-contains the final checkpoint of the approximately 180B-token run
-`prenorm640m-l3-hybrid500b23-180b-lr3e3-gbs256-mbs4-s4096`.
 It is a **base pretrained model** for text completion, not a chat/SFT model.
 Training code: [ykyaol7/ARLoop](https://github.com/ykyaol7/ARLoop).
-The loading implementation and file layout follow
-[ARLoop-1.3B](https://huggingface.co/osunlp/ARLoop-1.3B).
 ## Load and generate
-The included implementation uses PyTorch SDPA and standard pre-norm residual additions.
-No ARLoop source checkout, FLA, Triton or flash-attn installation is required.
 ```bash
 pip install 'transformers==5.22.0' safetensors
 # Install a PyTorch build suitable for your CPU or CUDA environment separately.
@@ -48,11 +42,6 @@ with torch.inference_mode():
 print(tokenizer.decode(output[0], skip_special_tokens=True))
 ```
-Keep the weights in FP32: the implementation preserves FP32 residuals and value embeddings with BF16
-autocast internally, matching the training model's mixed-precision policy.
-Total prompt plus generation length must not exceed 4096. For batches, the
-tokenizer uses left padding; pass its `attention_mask` to the model.
 ## Architecture
 | Property | Value |
@@ -68,50 +57,14 @@ tokenizer uses left padding; pass its `attention_mask` to the model.
 | Tokenizer vocabulary | 49152, SmolLM2 tokenizer used in training |
 | Output head | Tied to token embeddings |
-The inference checkpoint contains **582,804,096 unique parameters**. The training
-model also allocated an unused 75,497,472-parameter output head despite using
-tied embeddings; that unused weight is omitted here. Duplicate shared value
-embedding tensors are stored once. The exported weights retain their original
-FP32 values without lossy conversion.
-## Training checkpoint
-`main` contains the final checkpoint:
-| Property | Value |
-| --- | --- |
-| Original filename | `step-01373280-prenorm640m-l3-hybrid500b23-180b-lr3e3-gbs256-mbs4-s4096` |
-| Saved microbatch step | 1,373,280 |
-| Saved optimizer step | 171,660 |
-| Global batch size | 256 sequences |
-| Sequence length | 4096 tokens |
-| Training tokens | 179,998,556,160 |
-The token count is calculated from the saved optimizer step × 256 × 4096.
-Step labels in original filenames are microbatch steps, not optimizer steps.
-For reproducible experiments, pin the model repository's commit SHA.
-## Files and validation
-- `configuration_baseline.py`, `modeling_baseline.py`: standalone Transformers
-  loading code matching the native fixed-loop PreNorm architecture.
-- `config.json`, `generation_config.json`: architecture and generation defaults.
-- Two `model-*.safetensors` shards and `model.safetensors.index.json`: FP32 weights.
-- Tokenizer JSON, vocabulary, merges and special-token metadata.
-Export checks cover every tensor's shape, dtype and finiteness, equality of
-shared embeddings, and tokenizer vocabulary/merge consistency with the training
-run. Standalone loading, forward/loss and batched cached generation were tested
-with Python 3.12, PyTorch 2.6.0 (CPU) and Transformers 5.22.0, with imports of the
-training packages and FLA blocked. On the tested CPU input, logits exactly matched the native PreNorm model,
-and cached chunk inference exactly matched full forward inference. These are
-functional checks, not benchmark evaluations. GPU inference has not been tested
-for this release; different attention kernels may differ numerically in BF16.
-## License and attribution
-Model release: [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0).
-Attribution and MIT copyright/permission notices for adapted Parcae/ARLoop code
-are retained in `NOTICE`.
-The tokenizer files are copied from the training run's SmolLM2 tokenizer assets;
-these model weights were trained independently of SmolLM2.

 # ARLoop-baseline-0.6B
 This is the fixed-loop PreNorm baseline for ARLoop. This release
+contains the final checkpoint of the approximately 180B-token training run.
 It is a **base pretrained model** for text completion, not a chat/SFT model.
 Training code: [ykyaol7/ARLoop](https://github.com/ykyaol7/ARLoop).
 ## Load and generate
 ```bash
 pip install 'transformers==5.22.0' safetensors
 # Install a PyTorch build suitable for your CPU or CUDA environment separately.
 print(tokenizer.decode(output[0], skip_special_tokens=True))
 ```
 ## Architecture
 | Property | Value |
 | Tokenizer vocabulary | 49152, SmolLM2 tokenizer used in training |
 | Output head | Tied to token embeddings |
+## License
+Model release: [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0).
+## Training checkpoints
+`main` contains the final checkpoint.
+| Optimizer step | Training tokens | Revision |
+| ---: | ---: | --- |
+| 171,660 (final) | 179,998,556,160 | [main](https://huggingface.co/osunlp/ARLoop-baseline-0.6B/tree/main) |