diff --git a/README.md b/README.md
index 6dd1d14..a0a9850 100644
--- a/README.md
+++ b/README.md
@@ -3,13 +3,40 @@
 Reproduction of Anthropic's [Attribution Graphs](https://transformer-circuits.pub/2025/attribution-graphs/methods.html).
 
 ## Setup
+
+### Install UV
+
+**Option 1: Use `mise`** 
+Install mise: https://mise.jdx.dev/getting-started.html
+
+Environment setup can be done with:
+```sh
+mise install
 ```
-# install uv if you don't have it
-# curl -LsSf https://astral.sh/uv/install.sh | sh
+This will install `uv`.
+
+**Option 2: Direct install**
+
+```sh
+curl -LsSf https://astral.sh/uv/install.sh | sh
+```
+
+### Pull dependencies
+Once you have `uv` installed:
+
+```sh
 uv venv --seed
 uv sync
 ```
 
+### Huggingface Token
+To use some of the read restricted models, you'll need a registerd Huggingface Account and corresponding Read token. Use the following to pick it up:
+
+```sh
+export HF_TOKEN=<YOUR READ TOKEN>
+```
+
+
 ## Cache activations
 This step is not strictly necessary. Given the weights of a transcoder, we need to compute activations on a large dataset to be able to select maximum-activating examples.
 
@@ -25,7 +52,7 @@ uv run python cache.py $MODEL $TRANSCODER --num_gpus 1 $DATASET --hookpoints lay
 
 Alternatively, you can download the cache from Huggingface:
 
-```
+```sh
 cd results && git clone https://huggingface.co/nev/SmolLM2-CLT-135M-73k-k32-cache --depth=1
 ```
 
@@ -33,7 +60,7 @@ cd results && git clone https://huggingface.co/nev/SmolLM2-CLT-135M-73k-k32-cach
 
 We developed a gradio visualization for running attribution:
 
-```
+```sh
 uv run gradio serve.py
 ```
 
@@ -43,7 +70,7 @@ The web UI can be run from [a Colab notebook](https://colab.research.google.com/
 
 ## Run in CLI
 
-```
+```sh
 SCAN="smollm-v1"
 PROMPT_NAME="smollm-v1"
 PROMPT_TEXT="Michael Jordan plays the sport of"
@@ -55,12 +82,14 @@ uv run python -m attribute \
 --transcoder_path="nev/SmolLM2-CLT-135M-73k-k32" --model_name="HuggingFaceTB/SmolLM2-135M" \
 ```
 
-```
+```sh
 uv run python -m attribute
 cd attribution-graphs-frontend
 uv run python serve.py 9999
 ngrok http 9999
 ```
 
+Note: Ngrok now requires authentication
+
 ## TODOs
 - [ ] Cross-verify with Anthropic implementation
diff --git a/mise.toml b/mise.toml
new file mode 100644
index 0000000..07a2fe1
--- /dev/null
+++ b/mise.toml
@@ -0,0 +1,3 @@
+[tools]
+python = "3.13"
+uv = "latest"
diff --git a/pyproject.toml b/pyproject.toml
index 088ff35..87cc576 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -23,7 +23,7 @@ dependencies = [
     "transformers>=4.51.3",
     "gradio>=5.30.0",
     "neuronpedia>=1.0.20",
-    "datasets>=2.14.6",
+    "datasets==3.6.0",
 ]
 
 [tool.setuptools]