Instructions to use Omarrran/ks-byte-lm-spacebyte-transformers with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Omarrran/ks-byte-lm-spacebyte-transformers with Transformers:

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

pipe = pipeline("text-generation", model="Omarrran/ks-byte-lm-spacebyte-transformers", trust_remote_code=True)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Omarrran/ks-byte-lm-spacebyte-transformers", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Omarrran/ks-byte-lm-spacebyte-transformers with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Omarrran/ks-byte-lm-spacebyte-transformers

SGLang

How to use Omarrran/ks-byte-lm-spacebyte-transformers 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 "Omarrran/ks-byte-lm-spacebyte-transformers" \
    --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": "Omarrran/ks-byte-lm-spacebyte-transformers",
		"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 "Omarrran/ks-byte-lm-spacebyte-transformers" \
        --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": "Omarrran/ks-byte-lm-spacebyte-transformers",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Omarrran/ks-byte-lm-spacebyte-transformers with Docker Model Runner:
```
docker model run hf.co/Omarrran/ks-byte-lm-spacebyte-transformers
```

Omarrran commited on 30 days ago

Commit

1d2a4ca

verified ·

1 Parent(s): 18dbe3c

Refresh cleaned Transformers-compatible files

Browse files

Files changed (1) hide show

modeling_ksbyte.py +3 -3

modeling_ksbyte.py CHANGED Viewed

@@ -665,9 +665,9 @@ class KsByteForCausalLM(PreTrainedModel, GenerationMixin):
     supports_gradient_checkpointing = False
     _supports_cache_class = False
     _tied_weights_keys = ["model.lm_head.weight"]
-    # Newer Transformers releases (seen in Colab/PyTorch 2.8 stacks) access
-    # this public attribute during post-load tied-weight handling.
-    all_tied_weights_keys = ["model.lm_head.weight"]
     def __init__(self, config: KsByteConfig):
         super().__init__(config)

     supports_gradient_checkpointing = False
     _supports_cache_class = False
     _tied_weights_keys = ["model.lm_head.weight"]
+    # Some Transformers releases expect this to be a dict and call .keys();
+    # others only need membership. Keep it dict-shaped for Colab compatibility.
+    all_tied_weights_keys = {"model.lm_head.weight": "model.embed.weight"}
     def __init__(self, config: KsByteConfig):
         super().__init__(config)