Instructions to use AndrewThompson1233/maba-v1-101m-test with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AndrewThompson1233/maba-v1-101m-test with Transformers:

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

pipe = pipeline("text-generation", model="AndrewThompson1233/maba-v1-101m-test")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("AndrewThompson1233/maba-v1-101m-test", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AndrewThompson1233/maba-v1-101m-test with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AndrewThompson1233/maba-v1-101m-test"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AndrewThompson1233/maba-v1-101m-test",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/AndrewThompson1233/maba-v1-101m-test

SGLang

How to use AndrewThompson1233/maba-v1-101m-test 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 "AndrewThompson1233/maba-v1-101m-test" \
    --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": "AndrewThompson1233/maba-v1-101m-test",
		"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 "AndrewThompson1233/maba-v1-101m-test" \
        --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": "AndrewThompson1233/maba-v1-101m-test",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use AndrewThompson1233/maba-v1-101m-test with Docker Model Runner:
```
docker model run hf.co/AndrewThompson1233/maba-v1-101m-test
```

AndrewThompson1233 commited on 16 days ago

Commit

155d669

verified ·

1 Parent(s): 52841f9

release: v1.1 upgrade benchmark_results.json

Browse files

Files changed (1) hide show

benchmark_results.json +23 -1

benchmark_results.json CHANGED Viewed

@@ -1,5 +1,27 @@
 {
-  "Maba v1 (101M)": {
     "params": 101177984,
     "physical_blocks": 20,
     "effective_layers": 40,

 {
+  "Maba v1.1 (101M)": {
+    "params": 101177984,
+    "physical_blocks": 20,
+    "effective_layers": 40,
+    "gdn_ratio": "75%",
+    "gqa_ratio": "25%",
+    "val_loss_500": 5.8787,
+    "val_ppl_500": 357.34,
+    "arc_easy_acc": 26.8,
+    "hellaswag_acc": 24.0,
+    "tinystories_cloze_acc": 25.2,
+    "reasoning_margin": 0.2237,
+    "throughput_tok_s": 394.2,
+    "training_speed_tok_s": 2185,
+    "training_cluster_tok_s": 38400,
+    "kv_cache_4k_kb_runtime": 10240.0,
+    "kv_cache_4k_mb_runtime": 10.0,
+    "kv_cache_4k_kb_physical": 10240.0,
+    "kv_cache_4k_mb_physical": 10.0,
+    "kv_cache_reduction_runtime": "76.2%",
+    "kv_cache_reduction_physical": "76.2%"
+  },
+  "Maba v1.0 (Legacy)": {
     "params": 101177984,
     "physical_blocks": 20,
     "effective_layers": 40,