BenchmarkBot commited on
Commit
e0ef314
β€’
1 Parent(s): 9e3eaf4

fix naming

Browse files
Files changed (2) hide show
  1. app.py +3 -3
  2. src/assets/text_content.py +2 -2
app.py CHANGED
@@ -28,8 +28,8 @@ OPTIMUM_TOKEN = os.environ.get("OPTIMUM_TOKEN", None)
28
 
29
 
30
  ALL_COLUMNS_MAPPING = {
31
- "best_scored_model": "Best Scored LLM πŸ†",
32
- "model_type": "LLM Type πŸ€—",
33
  "weight_class": "Weight Class πŸ‹οΈ",
34
  #
35
  "backend.name": "Backend 🏭",
@@ -107,7 +107,7 @@ def get_benchmark_table(bench_df):
107
  # rename
108
  bench_df.rename(columns=ALL_COLUMNS_MAPPING, inplace=True)
109
  # transform
110
- bench_df["LLM Type πŸ€—"] = bench_df["LLM Type πŸ€—"].apply(process_model_type)
111
  bench_df["Weight Class πŸ‹οΈ"] = bench_df["Weight Class πŸ‹οΈ"].apply(
112
  process_weight_class
113
  )
 
28
 
29
 
30
  ALL_COLUMNS_MAPPING = {
31
+ "best_scored_model": "Best Scored Model πŸ†",
32
+ "model_type": "Model Type πŸ€—",
33
  "weight_class": "Weight Class πŸ‹οΈ",
34
  #
35
  "backend.name": "Backend 🏭",
 
107
  # rename
108
  bench_df.rename(columns=ALL_COLUMNS_MAPPING, inplace=True)
109
  # transform
110
+ bench_df["Model Type πŸ€—"] = bench_df["Model Type πŸ€—"].apply(process_model_type)
111
  bench_df["Weight Class πŸ‹οΈ"] = bench_df["Weight Class πŸ‹οΈ"].apply(
112
  process_weight_class
113
  )
src/assets/text_content.py CHANGED
@@ -12,9 +12,9 @@ A100_TEXT = """<h3>Single-GPU Benchmark (1xA100):</h3>
12
  <ul>
13
  <li>LLMs are evaluated on a singleton batch and genrating a thousand tokens.</li>
14
  <li>Peak memory is measured in MB during the first forward pass of the LLM (no warmup).</li>
15
- <li>Each pair of (LLM Type, Weight Class) is represented by the best scored LLM. This LLM is the one used for all the hardware/backend/optimization experiments.</li>
16
  <li>Score is the average evaluation score obtained from the <a href="https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard">πŸ€— Open LLM Leaderboard</a>.</li>
17
- <li>Ranking is based on the euclidean distance from "perfect LLM" (i.e. 0 latency and 100% accuracy).</li>
18
  </ul>
19
  """
20
 
 
12
  <ul>
13
  <li>LLMs are evaluated on a singleton batch and genrating a thousand tokens.</li>
14
  <li>Peak memory is measured in MB during the first forward pass of the LLM (no warmup).</li>
15
+ <li>Each pair of (Model Type, Weight Class) is represented by the best scored model. This LLM is the one used for all the hardware/backend/optimization experiments.</li>
16
  <li>Score is the average evaluation score obtained from the <a href="https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard">πŸ€— Open LLM Leaderboard</a>.</li>
17
+ <li>Ranking is based on the euclidean distance from the "Perfect LLM" (i.e. 0 latency and 100% accuracy).</li>
18
  </ul>
19
  """
20