docs: update card to bigsmall v3.14.1 (V10 streaming, version pin)

README.md

@@ -15,7 +15,7 @@ tags:
 ## Use it in 2 lines
 
 ```bash
-pip install bigsmall
+pip install "bigsmall>=3.14.1"
 ```
 
 ```python
@@ -55,10 +55,19 @@ model = BigSmallStreamingModel.from_pretrained(
 
 Uses up to ~12× less VRAM than standard loading by streaming layers on demand.
 
+## Stream straight from the Hub (no disk)
+
+```python
+import bigsmall
+state_dict = bigsmall.stream_from_hub("wpferrell/mistral-7b-instruct-bigsmall", device="cpu")
+```
+
+Decompresses directly from the HuggingFace CDN over HTTP range requests. With the default `cache=False`, no `.bs` file is ever written to disk (V10).
+
 ## Decompress to safetensors
 
 ```bash
-pip install bigsmall
+pip install "bigsmall>=3.14.1"
 bigsmall decompress wpferrell/mistral-7b-instruct-bigsmall -o mistral-7b-instruct-bigsmall/
 ```
 
@@ -69,7 +78,7 @@ This is a lossless-compressed copy of [mistralai/Mistral-7B-Instruct-v0.3](https
 ## Want to compress your own model?
 
 ```bash
-pip install bigsmall
+pip install "bigsmall>=3.14.1"
 bigsmall compress my-model/ -o my-model.bs
 ```
 
@@ -95,4 +104,4 @@ See [github.com/wpferrell/Bigsmall](https://github.com/wpferrell/Bigsmall) for t
 
 ## Requires
 
-`bigsmall >= 3.13.0` for the latest features. Earlier versions (>= 3.0.0) can still decode this model.
+`bigsmall >= 3.14.1` for the latest features. Earlier versions (>= 3.0.0) can still decode this model.