Upload README.md with huggingface_hub

README.md

@@ -13,3 +13,39 @@ model-index:
     - type: perplexity (BASIC)
       value: 32.05778110592746
 ---
+This is a d-Matrix functional reference of the OPT-125M model.
+The reference provides the following functional *configurations*:
+  Configuration | Explanation
+  :-- | :-- 
+  **`BASELINE`** | a reference functionally equivalent to the original model
+  **`BASIC`** | all linear algebraic operands quantized to `BFP16-64`, and all other operations transformed to approximated kernel simulations
+
+
+### Usage
+
+Install d-Matrix [Dmx_Compressor](https://github.com/d-matrix-ai/dmx-compressor) first.
+```sh
+pip install dmx_compressor
+pip install lm-eval
+```
+
+The following is an example model and its evaluation.
+
+```python
+from dmx.compressor import config_rules
+from dmx.compressor.modeling import DmxModel
+import lm_eval
+
+model_args = f"pretrained="d-matrix/d-matrix/opt-125m",revision="main",trust_remote_code=True"
+
+lm = lm_eval.api.registry.get_model("hf").create_from_arg_string(model_args, {"batch_size": 1})
+
+# Transform the model with DMX
+model = DmxModel.from_torch(lm._model)
+model.transform(model.dmx_config, *eval(f"config_rules.{dmx_config}"))  # Assign desired configuration to dmx_config, i.e. "BASELINE"
+lm._model = model  # Set the transformed model back
+
+task_dict = lm_eval.tasks.get_task_dict([task_name])  # Assign desired task to task_name, i.e. "wikitext"
+
+evaluation_results = lm_eval.evaluate(lm, task_dict)
+```