Upload README.md with huggingface_hub

README.md

@@ -1,40 +1,50 @@
 ---
-language:
-- en
-library_name: mlx
 license: apache-2.0
-tags:
-- mlx
+language: en
+library_name: mlx
 pipeline_tag: text-generation
 base_model: mlx-community/SmolLM-135M-Instruct-4bit
+tags:
+- mlx
+- 4-bit
+- apple-silicon
+- text-generation
 ---
 
-> **Renamed:** previously `harpertoken/harpertoken-apple-m1-model`. The old ID still redirects.
+> Previously published as `harpertoken/harpertoken-apple-m1-model`. The old identifier still redirects here.
 
-# harpertoken/harpertoken-apple-m1-model
+# tiny
 
-This model [harpertoken/harpertoken-apple-m1-model](https://huggingface.co/harpertoken/harpertoken-apple-m1-model) was
-converted to MLX format from [mlx-community/SmolLM-135M-Instruct-4bit](https://huggingface.co/mlx-community/SmolLM-135M-Instruct-4bit)
-using mlx-lm version **0.30.4**.
+A four-bit quantised conversion of [`mlx-community/SmolLM-135M-Instruct-4bit`](https://huggingface.co/mlx-community/SmolLM-135M-Instruct-4bit), produced with `mlx-lm` 0.30.4 and intended for Apple Silicon. The architecture is a Llama variant — thirty layers, 576 hidden dimensions, nine attention heads, three key-value heads, 49152-token vocabulary — with group-size 64 four-bit weights, occupying about 76 MB.
 
-## Use with mlx
+The `model.safetensors.index.json` alongside the weights is not a sharded checkpoint here; it maps all 694 tensors onto the single `model.safetensors` file that sits beside it.
 
-```bash
-pip install mlx-lm
-```
+Because the weights are quantised, `transformers` cannot load this checkpoint — it needs `mlx-lm`, which applies the group quantisation at load time. The instruction tuning lives upstream in SmolLM; this repository adds a format, nothing else. Given the size, expect it to be fast and largely incoherent on factual questions.
+
+## Usage
 
 ```python
 from mlx_lm import load, generate
 
-model, tokenizer = load("harpertoken/harpertoken-apple-m1-model")
+model, tokenizer = load("harpertoken/tiny")
 
 prompt = "hello"
-
 if tokenizer.chat_template is not None:
-    messages = [{"role": "user", "content": prompt}]
     prompt = tokenizer.apply_chat_template(
-        messages, add_generation_prompt=True, return_dict=False,
+        [{"role": "user", "content": prompt}],
+        add_generation_prompt=True,
+        return_dict=False,
     )
 
-response = generate(model, tokenizer, prompt=prompt, verbose=True)
+print(generate(model, tokenizer, prompt=prompt, max_tokens=100))
 ```
+
+Install with `pip install mlx-lm`. A `chat_template.jinja` is present, so the template branch above will be taken.
+
+## Limitations
+
+This is a redistribution of someone else's work in a different numeric format. There is no training, evaluation or adaptation to report, and if you want SmolLM-135M-Instruct you should use the upstream repository, which is not tied to MLX and can be converted to other formats. Quantisation to four bits costs some accuracy even before the underlying 135M-parameter capacity does; I have not measured the loss and so make no claim about it.
+
+## Attribution
+
+SmolLM follows Allouji et al., *The SmolLM2 Family of Small Language Models* (2025).