Spaces:
Running
Running
| import gradio as gr | |
| import pandas as pd | |
| import plotly.express as px | |
| ATTN_DATA = [ | |
| # open llm | |
| "Model π€", | |
| "Experiment π§ͺ", | |
| "Params (B)", | |
| "Architecture ποΈ", | |
| "Open LLM Score (%)", | |
| # deployment settings | |
| "Backend π", | |
| "Quantization ποΈ", | |
| "Precision π₯", | |
| "Attention ποΈ", | |
| "Kernel βοΈ", | |
| # primary measurements | |
| "Prefill (s)", | |
| "Decode (tokens/s)", | |
| # speedups | |
| "Prefill Speedup (%)", | |
| "Decode Speedup (%)", | |
| ] | |
| def get_attn_df(open_llm_perf_df): | |
| copy_df = open_llm_perf_df.copy() | |
| copy_df["Quantization & Kernel"] = ( | |
| copy_df["Quantization ποΈ"] + " & " + copy_df["Kernel βοΈ"] | |
| ) | |
| eager_df = copy_df[(copy_df["Attention ποΈ"] == "Eager")] | |
| sdpa_df = copy_df[(copy_df["Attention ποΈ"] == "SDPA")] | |
| fa2_df = copy_df[(copy_df["Attention ποΈ"] == "FAv2")] | |
| sdpa_df = pd.merge( | |
| eager_df, | |
| sdpa_df, | |
| on=["Model π€", "Quantization & Kernel"], | |
| suffixes=["", " other"], | |
| ) | |
| fa2_df = pd.merge( | |
| eager_df, | |
| fa2_df, | |
| on=["Model π€", "Quantization & Kernel"], | |
| suffixes=["", " other"], | |
| ) | |
| attn_df = pd.concat([sdpa_df, fa2_df]) | |
| # compute speedups | |
| attn_df["Prefill Speedup (%)"] = ( | |
| (attn_df["Prefill (s)"] / attn_df["Prefill (s) other"]) * 100 | |
| ).round(2) - 100 | |
| attn_df["Decode Speedup (%)"] = ( | |
| (attn_df["Decode (tokens/s) other"] / attn_df["Decode (tokens/s)"]) * 100 | |
| ).round(2) - 100 | |
| return attn_df | |
| def get_attn_prefill_fig(open_llm_perf_df): | |
| attn_df = get_attn_df(open_llm_perf_df) | |
| # plot | |
| prefill_fig = px.box( | |
| attn_df, | |
| x="Architecture ποΈ", | |
| y="Prefill Speedup (%)", | |
| color_discrete_sequence=px.colors.qualitative.Light24, | |
| custom_data=ATTN_DATA, | |
| color="Attention ποΈ other", | |
| points="all", | |
| ) | |
| # add hover data | |
| prefill_fig.update_traces( | |
| hovertemplate="<br>".join( | |
| [ | |
| f"<b>{column}:</b> %{{customdata[{i}]}}" | |
| for i, column in enumerate(ATTN_DATA) | |
| ] | |
| ) | |
| ) | |
| # add layout | |
| prefill_fig.update_layout( | |
| title={ | |
| "text": "Prefill Speedup per Architecture, Compared To Eager Attention", | |
| "xanchor": "center", | |
| "yanchor": "top", | |
| "y": 0.95, | |
| "x": 0.5, | |
| }, | |
| yaxis_title="Prefill Speedup (%)", | |
| xaxis_title="LLM Architecture", | |
| legend_title="Attention", | |
| width=1200, | |
| height=600, | |
| ) | |
| return prefill_fig | |
| def get_attn_decode_fig(open_llm_perf_df): | |
| attn_df = get_attn_df(open_llm_perf_df) | |
| print(len(attn_df)) | |
| # plot | |
| decode_fig = px.box( | |
| attn_df, | |
| x="Architecture ποΈ", | |
| y="Decode Speedup (%)", | |
| color_discrete_sequence=px.colors.qualitative.Light24, | |
| custom_data=ATTN_DATA, | |
| color="Attention ποΈ other", | |
| points="all", | |
| ) | |
| # add hover data | |
| decode_fig.update_traces( | |
| hovertemplate="<br>".join( | |
| [ | |
| f"<b>{column}:</b> %{{customdata[{i}]}}" | |
| for i, column in enumerate(ATTN_DATA) | |
| ] | |
| ) | |
| ) | |
| # add layout | |
| decode_fig.update_layout( | |
| title={ | |
| "text": "Decode Speedup per Architecture, Compared To Eager Attention", | |
| "xanchor": "center", | |
| "yanchor": "top", | |
| "y": 0.95, | |
| "x": 0.5, | |
| }, | |
| yaxis_title="Decode Speedup (%)", | |
| xaxis_title="LLM Architecture", | |
| legend_title="Attention", | |
| width=1200, | |
| height=600, | |
| ) | |
| return decode_fig | |
| def create_attn_plots(open_llm_perf_df): | |
| # descriptive text | |
| gr.HTML("π Hover over the points π for additional information.", elem_id="text") | |
| # get figures | |
| prefill_fig = get_attn_prefill_fig(open_llm_perf_df) | |
| decode_fig = get_attn_decode_fig(open_llm_perf_df) | |
| # create plots | |
| prefill_plot = gr.components.Plot( | |
| value=prefill_fig, elem_id="plot", show_label=False | |
| ) | |
| decode_plot = gr.components.Plot(value=decode_fig, elem_id="plot", show_label=False) | |
| return prefill_plot, decode_plot | |