smol changes

app.py

@@ -68,6 +68,8 @@ model_info = "N/A"
 ### Conversion factors
 kg_per_mile = 0.348
 embodied_conversion_factor = 0.0289
+miles_km_conversion_factor = 0.62137119
+
 
 if "is_shared" not in st.session_state:
     st.session_state["is_shared"] = False
@@ -111,8 +113,8 @@ with st.expander("Calculate the dynamic emissions of your model"):
                         ' [Electricity Map](https://app.electricitymaps.com/) to get this information.')
     dynamic_emissions = round(gpu_tdp * float(num_gpus)*training_time * carbon_intensity/1000000)
     st.metric(label="Dynamic emissions", value=str(dynamic_emissions)+' kilograms of CO2eq')
-    st.info('This is roughly equivalent to '+ str(round(dynamic_emissions/kg_per_mile,1)) + ' miles driven in an average US car'
-    ' produced in 2021. [(Source: energy.gov)](https://www.energy.gov/eere/vehicles/articles/fotw-1223-january-31-2022-average-carbon-dioxide-emissions-2021-model-year)')
+    st.info('This is roughly equivalent to '+ str(round(dynamic_emissions/kg_per_mile)) + ' miles ('+ str(round(dynamic_emissions/kg_per_mile/miles_km_conversion_factor)) + ' km) driven in an average US car'
+    '. [(Source: energy.gov)](https://www.energy.gov/eere/vehicles/articles/fotw-1223-january-31-2022-average-carbon-dioxide-emissions-2021-model-year)')
 
 st.markdown('### Experimental Emissions 👩‍🔬')
 st.markdown('##### These are the emissions produced by generating the electricity necessary for powering the experiments and tests needed to pick your final model architecture '
@@ -199,7 +201,18 @@ with buttoncol2:
 
 st.markdown('### Methodology')
 with st.expander("More information about our Methodology"):
-    st.markdown('Building on the work of the [ML CO2 Calculator](https://mlco2.github.io/impact/), this tool allows you to consider'
-                ' other aspects of your model\'s carbon footprint based on the LCA methodology.')
-    st.markdown('We considered all of these aspects when calculating the CO2 emissions of BLOOM 🌸, a 176-billion parameter language model [(see our preprint!)](https://arxiv.org/abs/2211.02001)'')')
-    st.image('images/LCA_CO2.png', caption='The LCA methodology - the parts in green are those we focus on.')
+    st.markdown('While there is no universally-accepted approach for assessing the environmental impacts of ML models,'
+    ' we strive towards adopting the widely-used Life Cycle Assessment (LCA) methodology, which aims to cover all stages '
+    'of the life cycle of a product or process. We focus on the steps for which we do have sufficient information, which range from manufacturing the equipment used '
+    ' for training the model to energy consumption (in green in the figure below).')
+    st.image('images/LCA_CO2.png', caption='The LCA methodology applied to AI models.')
+    st.info('We are using all of the available sources of information that we can, and will update the tool as new sources'
+    ' of information are published.')
+    st.markdown('#### Dynamic and Experimental Emissions')
+    st.markdown('These are calculated by multiplying the number of GPU hours used by the thermal design power (TDP) of those GPUs and the carbon intensity of the energy grid used to power the hardware.')
+    st.markdown('#### Datacenter Emissions')
+    st.markdown('These are calculated by multiplying the dynamic emissions by the PUE (Power Usage Effectiveness) of the datacenter used for running the code.')
+    st.markdown('#### Embodied Emissions')
+    st.markdown('These are calculated by dividing the amount of emissions produced during the manufacturing process by the time of use based on a linearly ammortised emissions conversion factor (in our case, 0,0289 [(source)](https://docs.google.com/spreadsheets/d/1DqYgQnEDLQVQm5acMAhLgHLD8xXCG9BIrk-_Nv6jF3k/)).')
+    st.markdown('#### Example of application')
+    st.markdown('For an example this methodology, check out the [BLOOM 🌸 model CO2 paper](https://arxiv.org/abs/2211.02001)'')')