Fix small typos in README.md

README.md

@@ -10,7 +10,7 @@ Chroma is a generative model for designing proteins **programmatically**.
 
 Protein space is complex and hard to navigate. With Chroma, protein design problems are represented in terms of [composable building blocks](#conditioners) from which diverse, [all-atom protein structures can be automatically generated](#sampling). As a joint model of structure and sequence, Chroma can also be used for common protein modeling tasks such as [generating sequences given backbones](#design), packing side-chains, and scoring designs. 
 
-We provide protein conditioners for a variety of constraints, including substructure, symmetry, shape, and neural-network predictions of some protein classes and annotations. We also provide an API for [creating your own conditioners](#conditioners-api) in a few lines of code.
+We provide protein conditioners for a variety of constraints, including substructure, symmetry, shape, and neural network predictions of some protein classes and annotations. We also provide an API for [creating your own conditioners](#conditioners-api) in a few lines of code.
 
 Internally, Chroma uses diffusion modeling, equivariant graph neural networks, and conditional random fields to efficiently sample all-atom structures with a complexity that is sub-quadratic in the number of residues. It can generate large complexes in a few minutes on a commodity GPU. You can read more about Chroma, including biophysical and crystallographic validation of some early designs, in our paper, [*Illuminating protein space with a programmable generative model*. Nature 2023](https://doi.org/10.1038/s41586-023-06728-8). 
 
@@ -84,7 +84,7 @@ protein.to("sample-C3.cif")
 Because compositions of conditioners are conditioners, even relatively complex design problems can follow this basic usage pattern. See the [demo notebooks](#get-started) and docstrings for more information on hyperparameters, conditioners, and starting points.
 
 ## Design
-**Robust design**. Chroma is a joint model of sequence and structure that uses a common graph neural network base architecture to parameterize both backbone generation and conditional sequence and sidechain generation. These sequence and sidechain decoders are *diffusion-aware* in the sense that they have been trained to predict sequence and side chain not just for natural structures at diffusion time $t=0$ but also on noisy structures at all diffusion times $t \in [0,1]$. As a result, the $t$ hyperpameter of the design network provides a kind of tunable robustness via **diffusion augmentation** in we trade off between how much the model attempts to design the backbone *exactly* as specified (e.g. $t=0.0$) versus *robust* design within a small neighborhood of nearby backbone conformations (e.g. $t=0.5$).
+**Robust design**. Chroma is a joint model of sequence and structure that uses a common graph neural network base architecture to parameterize both backbone generation and conditional sequence and sidechain generation. These sequence and sidechain decoders are *diffusion-aware* in the sense that they have been trained to predict sequence and side chain not just for natural structures at diffusion time $t=0$ but also on noisy structures at all diffusion times $t \in [0,1]$. As a result, the $t$ hyperparameter of the design network provides a kind of tunable robustness via **diffusion augmentation** in we trade-off between how much the model attempts to design the backbone *exactly* as specified (e.g. $t=0.0$) versus *robust* design within a small neighborhood of nearby backbone conformations (e.g. $t=0.5$).
 
 While all results presented in the Chroma [publication](https://doi.org/10.1038/s41586-023-06728-8) were done with **exact design** at $t=0.0$, we have found **robust design** at times near $t=0.5$ frequently improves one-shot refolding while incurring only minor, often Ångstrom-scale, relaxation adjustments to target backbones. When we compare the performance of these two design modes on our set of 50,000 unconditional backbones that were analyzed in the paper, we see very large improvements in refolding across both [AlphaFold](https://github.com/google-deepmind/alphafold) and [ESMFold](https://github.com/facebookresearch/esm) that stratifies well across protein length, percent helicity, or similarity to a known structure (See Chroma [Supplementary Figure 14](https://doi.org/10.1038/s41586-023-06728-8) for further context).
 
@@ -173,7 +173,7 @@ class Conditioner(torch.nn.Module):
         U_update = U + update_energy(X, C, t)
         return X_update, C_update, O, U_update, t
 ```
- Roughly speaking, `Conditioner`s are composable by construction because their input and output type signatures are matched (i.e. they are an endomorphism). So we also simply build conditioners from conditioners by "stacking" them much as we would with traditional neural network layer developemnt. With the final `Conditioner` as an input, `Chroma.sample()` will then leverage Pytorch's automatic differentiation facilities to automaticallly furnish a diffusion-annealed MCMC sampling algorithm to sample with this conditioner (We note this isn't magic and taking care to scale and parameterize appropriately is [important](#note-on-conditioners)).
+ Roughly speaking, `Conditioner`s are composable by construction because their input and output type signatures are matched (i.e. they are an endomorphism). So we also simply build conditioners from conditioners by "stacking" them much as we would with traditional neural network layer development. With the final `Conditioner` as an input, `Chroma.sample()` will then leverage Pytorch's automatic differentiation facilities to automatically furnish a diffusion-annealed MCMC sampling algorithm to sample with this conditioner (We note this isn't magic and taking care to scale and parameterize appropriately is [important](#note-on-conditioners)).
 
 ##### A minimal Conditioner: 2D lattice symmetry
 The code snippet below shows how in a few lines of code we can add a conditioner that stipulates the generation of a 2D crystal-like object, where generated proteins are arrayed in an `M x N` rectangular lattice.
@@ -218,7 +218,7 @@ protein.to_CIF("lattice_protein.cif")
 
 #### Note on Conditioners
 
-An attractive aspect of this conditioner framework is that it is very general, enabling both constraints (which involve operations on $x$) and restraints (which amount to changes to $U$). At the same time, generation under restraints can still be (and often is) challenging, as the resulting effective energy landscape can become arbitrarily rugged and difficult to integrate. We therefore advise caution when using and developing new conditioners or conditioner combinations. We find that inspecting diffusition trajectories (including unconstrained and denoised trajectories, $\hat{x}_t$ and $\tilde{x}_t$) can be a good tool for identifying integration challenges and defining either better conditioner forms or better sampling regimes.
+An attractive aspect of this conditioner framework is that it is very general, enabling both constraints (which involve operations on $x$) and restraints (which amount to changes to $U$). At the same time, generation under restraints can still be (and often is) challenging, as the resulting effective energy landscape can become arbitrarily rugged and difficult to integrate. We therefore advise caution when using and developing new conditioners or conditioner combinations. We find that inspecting diffusion trajectories (including unconstrained and denoised trajectories, $\hat{x}_t$ and $\tilde{x}_t$) can be a good tool for identifying integration challenges and defining either better conditioner forms or better sampling regimes.
 
 ## Citing Chroma
 
@@ -240,7 +240,7 @@ J. B. Ingraham, M. Baranov, Z. Costello, K. W. Barber, W. Wang, A. Ismail, V. Fr
 ```
 
 ## Acknowledgements
-The Chroma codebase is the work of many contributers at Generate Biomedicines. We would like to acknowledge: Ahmed Ismail, Alan Witmer, Alex Ramos, Alexander Bock, Ameya Harmalkar, Brinda Monian, Craig Mackenzie, Dan Luu, David Moore, Frank Oplinger, Fritz Obermeyer, George Kent-Scheller, Gevorg Grigoryan, Jacob Feala, James Lucas, Jenhan Tao, John Ingraham, Martin Jankowiak, Max Baranov, Meghan Franklin, Mick Ward, Rudraksh Tuwani, Ryan Nelson, Shan Tie, Vincent Frappier, Vincent Xue, William Wolfe-McGuire, Wujie Wang, Zak Costello, Zander Harteveld.
+The Chroma codebase is the work of many contributors at Generate Biomedicines. We would like to acknowledge: Ahmed Ismail, Alan Witmer, Alex Ramos, Alexander Bock, Ameya Harmalkar, Brinda Monian, Craig Mackenzie, Dan Luu, David Moore, Frank Oplinger, Fritz Obermeyer, George Kent-Scheller, Gevorg Grigoryan, Jacob Feala, James Lucas, Jenhan Tao, John Ingraham, Martin Jankowiak, Max Baranov, Meghan Franklin, Mick Ward, Rudraksh Tuwani, Ryan Nelson, Shan Tie, Vincent Frappier, Vincent Xue, William Wolfe-McGuire, Wujie Wang, Zak Costello, Zander Harteveld.
 
 ## License