File size: 12,419 Bytes
e7abd9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e3acf44
e7abd9e
 
 
3adbd07
cf713d3
e7abd9e
 
077602c
3adbd07
e7abd9e
 
 
04f06ea
3fc563a
 
 
04f06ea
cf713d3
3fc563a
 
077602c
3fc563a
 
 
 
04f06ea
62aa434
 
 
 
cf713d3
 
 
 
 
 
 
 
04f06ea
cf713d3
 
 
04f06ea
cf713d3
 
 
 
 
 
 
 
04f06ea
cf713d3
 
 
04f06ea
cf713d3
62aa434
 
 
 
 
 
 
e7abd9e
 
 
 
 
014d36a
 
 
 
 
 
e7abd9e
 
 
 
 
 
014d36a
 
 
e7abd9e
 
 
 
 
 
 
 
 
 
014d36a
 
 
 
e7abd9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
014d36a
e7abd9e
014d36a
 
 
 
e7abd9e
 
 
014d36a
 
e7abd9e
014d36a
 
 
e7abd9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a5b3d73
e7abd9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8e899eb
e7abd9e
 
8e899eb
 
 
 
 
014d36a
e7abd9e
014d36a
e7abd9e
 
014d36a
e7abd9e
8e899eb
e7abd9e
 
8e899eb
e7abd9e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
import { Box, Typography } from "@mui/material";

const createTooltipContent = (title, items) => (
  <Box sx={{ maxWidth: 400 }}>
    <Typography variant="body2" paragraph sx={{ mb: 1, color: "inherit" }}>
      {title}
    </Typography>
    <Box component="ul" sx={{ m: 0, pl: 2 }}>
      {items.map(({ label, description, subItems }, index) => (
        <li key={index}>
          <Typography variant="body2" sx={{ mb: 0.5, color: "inherit" }}>
            <b>{label}</b>: {description}
            {subItems && (
              <Box component="ul" sx={{ mt: 0.5, mb: 1 }}>
                {subItems.map((item, subIndex) => (
                  <li key={subIndex}>
                    <Typography variant="body2" sx={{ color: "inherit" }}>
                      {item}
                    </Typography>
                  </li>
                ))}
              </Box>
            )}
          </Typography>
        </li>
      ))}
    </Box>
  </Box>
);

export const COLUMN_TOOLTIPS = {
  AVERAGE: createTooltipContent("Average score across all benchmarks:", [
    {
      label: "Calculation",
      description: "Weighted average of normalized scores from all benchmarks",
      subItems: [
        "Each benchmark is normalized to a 0-100 scale",
        "All normalised benchmarks are then averaged together",
      ],
    },
  ]),

  MULTIFIN: createTooltipContent("Multilingual Financial NLP (MultiFin):", [
    {
      label: "Purpose",
      description:
        "Tests model's ability to understand real-world financial article headlines",
      subItems: ["Language Understanding", "Topic Classification"],
    },
    {
      label: "Scoring: ACC Norm",
      description: "Was the correct choice selected among the options.",
    },
  ]),

  QA: createTooltipContent("Greek Financial Question Answering (QA):", [
    {
      label: "Purpose",
      description:
        "Tests model's ability to understand real-world Greek financial exam questions",
      subItems: ["Language Understanding", "Question Answering"],
    },
    {
      label: "Scoring: ACC Norm",
      description: "Was the correct choice selected among the options.",
    },
  ]),

  FNS: createTooltipContent("Financial Narrative Summarization (FNS):", [
    {
      label: "Purpose",
      description:
        "Tests model's ability to summarize real-world financial annual reports",
      subItems: ["Language Understanding", "Abstractive Summarization"],
    },
    {
      label: "Scoring: Rouge1",
      description: "Was the overlap of unigrams (each word) between the predicted and reference summaries.",
    },
  ]),

  FinNum: createTooltipContent("Financial Number Entity Recognition (FinNum):", [
    {
      label: "Purpose",
      description:
        "Tests model's ability to recognize numbers from financial texts",
      subItems: ["Language Understanding", "Named Entity Recognition"],
    },
    {
      label: "Scoring: Rouge1",
      description: "Was the overlap of unigrams (each word) between the predicted and reference summaries.",
    },
  ]),

  FinText: createTooltipContent("Financial Named Entity Recognition (FinText):", [
    {
      label: "Purpose",
      description:
        "Tests model's ability to recognize financial entities, i.e, person, location, and organization",
      subItems: ["Language Understanding", "Named Entity Recognition"],
    },
    {
      label: "Scoring: Rouge1",
      description: "Was the overlap of unigrams (each word) between the predicted and reference summaries.",
    },
  ]),

  ARCHITECTURE: createTooltipContent("Model Architecture Information:", [
    {
      label: "Definition",
      description: "The fundamental structure and design of the model",
      subItems: [
        "Pretrained: Foundational models, initially trained on large datasets without task-specific tuning, serving as a versatile base for further development.",
        "Continuously Pretrained: Base models trained with a data mix evolving as the model is trained, with the addition of specialized data during the last training steps.",
        "Fine-tuned: Base models, fine-tuned on specialised domain data (legal, medical, ...), and optimized for particular tasks.",
        "Chat: Models fine-tuned with IFT, RLHF, DPO, and other techniques, to handle conversational contexts effectively.",
        "Merged: Combining multiple models through weights averaging or similar methods.",
        "Multimodal: Models which can handle several modalities (text & image/audio/video/...). We only evaluate the text capabilities.",
      ],
    },
    {
      label: "Impact",
      description: "How architecture affects model capabilities",
      subItems: [
        "Base models are expected to perform less well on instruction following evaluations, like IFEval.",
        "Fine-tuned and chat models can be more verbose and more chatty than base models.",
        "Merged models tend to exhibit good performance on benchmarks, which do not translate to real-world situations.",
      ],
    },
  ]),

  PRECISION: createTooltipContent("Numerical Precision Format:", [
    {
      label: "Overview",
      description:
        "Data format used to store model weights and perform computations",
      subItems: [
        "bfloat16: Half precision (Brain Float format), good for stability",
        "float16: Half precision",
        "8bit/4bit: Quantized formats, for efficiency",
        "GPTQ/AWQ: Quantized methods",
      ],
    },
    {
      label: "Impact",
      description: "How precision affects model deployment",
      subItems: [
        "Higher precision = better accuracy but more memory usage",
        "Lower precision = faster inference and smaller size",
        "Trade-off between model quality and resource usage",
      ],
    },
  ]),

  FLAGS: createTooltipContent("Model Flags and Special Features:", [
    {
      label: "Filters",
      subItems: [
        "Mixture of Expert: Uses a MoE architecture",
        "Merged models: Created by averaging other models",
        "Contaminated: Flagged by users from the community for (possibly accidental) cheating",
        "Unavailable: No longer on the hub (private, deleted) or missing a license tag",
      ],
    },
    {
      label: "Purpose",
      description: "Why do people want to hide these models?",
      subItems: [
        "Mixture of Experts: These models can be too parameter heavy",
        "Merged models: Performance on benchmarks tend to be inflated compared to real life usage",
        "Contaminated: Performance on benchmarks is inflated and not reflecting real life usage",
      ],
    },
  ]),

  PARAMETERS: createTooltipContent("Model Parameters:", [
    {
      label: "Measurement",
      description: "Total number of trainable parameters in billions",
      subItems: [
        "Indicates model capacity and complexity",
        "Correlates with computational requirements",
        "Influences memory usage and inference speed",
      ],
    },
  ]),

  LICENSE: createTooltipContent("Model License Information:", [
    {
      label: "Importance",
      description: "Legal terms governing model usage and distribution",
      subItems: [
        "Commercial vs non-commercial use",
        "Attribution requirements",
        "Modification and redistribution rights",
        "Liability and warranty terms",
      ],
    },
  ]),

  CO2_COST: createTooltipContent("Carbon Dioxide Emissions:", [
    {
      label: "What is it?",
      description: "CO₂ emissions of the model evaluation ",
      subItems: [
        "Only focuses on model inference for our specific setup",
        "Considers data center location and energy mix",
        "Allows equivalent comparision of models on our use case",
      ],
    },
    {
      label: "Why it matters",
      description: "Environmental impact of AI model training",
      subItems: [
        "Large models can have significant carbon footprints",
        "Helps make informed choices about model selection",
      ],
    },
    {
      label: "Learn more",
      description:
        "For detailed information about our CO₂ calculation methodology, visit:",
      subItems: [
        <a
          href="https://huggingface.co/docs/leaderboards/open_llm_leaderboard/emissions"
          target="_blank"
          rel="noopener noreferrer"
          style={{ color: "#90caf9" }}
        >
          Carbon Emissions Documentation ↗
        </a>,
      ],
    },
  ]),
};

export const UI_TOOLTIPS = {
  COLUMN_SELECTOR: "Choose which columns to display in the table",
  DISPLAY_OPTIONS: createTooltipContent("Table Display Options", [
    {
      label: "Overview",
      description: "Configure how the table displays data and information",
      subItems: [
        "Row size and layout",
        "Score display format",
        "Ranking calculation",
        "Average score computation",
      ],
    },
  ]),
  SEARCH_BAR: createTooltipContent("Advanced Model Search", [
    {
      label: "Name Search",
      description: "Search directly by model name",
      subItems: [
        "Supports regular expressions (e.g., ^mistral.*7b)",
        "Case sensitive",
      ],
    },
    {
      label: "Field Search",
      description: "Use @field:value syntax for precise filtering",
      subItems: [
        "@architecture:llama - Filter by architecture",
        "@license:mit - Filter by license",
        "@precision:float16 - Filter by precision",
        "@type:chat - Filter by model type",
      ],
    },
    {
      label: "Multiple Searches",
      description: "Combine multiple criteria using semicolons",
      subItems: [
        "meta @license:mit; @architecture:llama",
        "^mistral.*7b; @precision:float16",
      ],
    },
  ]),
  QUICK_FILTERS: createTooltipContent(
    "Filter models based on their size and applicable hardware:",
    [
      {
        label: "Edge devices (Up to 3BB)",
        description:
          "Efficient models for edge devices, optimized for blazing fast inference.",
      },
      {
        label: "Smol Models (3B-7B)",
        description:
          "Efficient models for consumer hardware, optimized for fast inference.",
      },
      {
        label: "Mid-range models (7B-65B)",
        description:
          "A bit of everything here, with overall balanced performance and resource usage around 30B.",
      },
      {
        label: "GPU-rich models (65B+)",
        description:
          "State-of-the-art performance for complex tasks, requires significant computing power.",
      },
      {
        label: "Official Providers",
        description:
          "Models directly maintained by their original creators, ensuring reliability and up-to-date performance.",
      },
    ]
  ),
  ROW_SIZE: {
    title: "Row Size",
    description:
      "Adjust the height of table rows. Compact is ideal for viewing more data at once, while Large provides better readability and touch targets.",
  },
  SCORE_DISPLAY: {
    title: "Score Display",
    description:
      "Choose between normalized scores (0-100% scale for easy comparison) or raw scores (actual benchmark results). Normalized scores help compare performance across different benchmarks, while raw scores show actual benchmark outputs.",
  },
  RANKING_MODE: {
    title: "Ranking Mode",
    description:
      "Choose between static ranking (original position in the full leaderboard) or dynamic ranking (position based on current filters and sorting).",
  },
  AVERAGE_SCORE: {
    title: "Average Score Calculation",
    description:
      "Define how the average score is calculated. 'All Scores' uses all benchmarks, while 'Visible Only' calculates the average using only the visible benchmark columns.",
  },
};

export const getTooltipStyle = {};

export const TABLE_TOOLTIPS = {
  HUB_LINK: (modelName) => `View ${modelName} on Hugging Face Hub`,
  EVAL_RESULTS: (modelName) =>
    `View detailed evaluation results for ${modelName}`,
  POSITION_CHANGE: (change) =>
    `${Math.abs(change)} position${Math.abs(change) > 1 ? "s" : ""} ${
      change > 0 ? "up" : "down"
    }`,
  METADATA: {
    TYPE: (type) => type || "-",
    ARCHITECTURE: (arch) => arch || "-",
    PRECISION: (precision) => precision || "-",
    LICENSE: (license) => license || "-",
    UPLOAD_DATE: (date) => date || "-",
    SUBMISSION_DATE: (date) => date || "-",
    BASE_MODEL: (model) => model || "-",
  },
};