Updated readme.

README.md

@@ -88,8 +88,8 @@ print(tokenizer.decode(generated_ids[0], skip_special_tokens=True))
 
 - `reference_rope`: If set to `True`, RoPE implementation implemented in the original paper is used. This is the case for model pretrained in this repo. If set to `False`, RoPE implementation from the Huggingface Transformers library is used.
 
-- `output_pldrllm_attentions=True` returns the deductive outputs and learnable parameters of power law graph attention module as tuple containing:
-the output of the residual metric learner (metric tensor, $\textbf{A}$), output ($\textbf{A}_{LM}$) after application of iSwiGLU on metric tensor, learned exponents of potential tensor, learned weights for energy-curvature tensor, learned bias for energy-curvature tensor, energy-curvature tensor ($\textbf{G}_{LM}$), and attention weights.
+- `output_pldr_attentions=True` returns the deductive outputs and learnable parameters of power law graph attention module as tuple containing:
+the output of the residual metric learner (metric tensor, **A**), output (**A<sub>LM</sub>**) after application of iSwiGLU on metric tensor, learned exponents of potential tensor, learned weights for energy-curvature tensor, learned bias for energy-curvature tensor, energy-curvature tensor (**G<sub>LM</sub>**), and attention weights.
 
 See config.json for other model configuration details.
 
@@ -105,6 +105,7 @@ We also have a fork of transformers library with PLDR-LLM model support for futu
 ```python
       git clone https://github.com/burcgokden/transformers
       cd transformers
+      git checkout add_PLDR_LLM
       pip install -e ".[dev]"
 ```
 - Static cache is not supported for models with `custom_G_type=None`.