orionweller commited on
Commit
f1fa713
1 Parent(s): 83fe3f0

add_followir_tab (#102)

Browse files

- add instruction following (cf7ddc6fa21dbe43cd4c5f806946100e72ba8b72)
- update (0d0563c24895d35047cbd4018d9f86afb7e8a239)
- merge in main (aeb9d6091824165ded58f9a5f3b230a4434986b5)
- minor cleanup (b5c28bdf082216a290351c60305b599a952c7a73)
- add bi-encoder button (77cc9e7a65257c5af5784bb60a3dac2073e7fe05)

Files changed (4) hide show
  1. EXTERNAL_MODEL_RESULTS.json +0 -0
  2. app.py +39 -7
  3. config.yaml +25 -0
  4. model_meta.yaml +134 -0
EXTERNAL_MODEL_RESULTS.json CHANGED
The diff for this file is too large to render. See raw diff
 
app.py CHANGED
@@ -17,6 +17,11 @@ TASKS_CONFIG = LEADERBOARD_CONFIG["tasks"]
17
  BOARDS_CONFIG = LEADERBOARD_CONFIG["boards"]
18
 
19
  TASKS = list(TASKS_CONFIG.keys())
 
 
 
 
 
20
 
21
  TASK_TO_METRIC = {k:v["metric"] for k,v in TASKS_CONFIG.items()}
22
 
@@ -34,18 +39,30 @@ EXTERNAL_MODEL_TO_DIM = {k: v["dim"] for k,v in MODEL_META["model_meta"].items()
34
  EXTERNAL_MODEL_TO_SEQLEN = {k: v["seq_len"] for k,v in MODEL_META["model_meta"].items() if v.get("seq_len", False)}
35
  EXTERNAL_MODEL_TO_SIZE = {k: v["size"] for k,v in MODEL_META["model_meta"].items() if v.get("size", False)}
36
  PROPRIETARY_MODELS = {k for k,v in MODEL_META["model_meta"].items() if v.get("is_proprietary", False)}
 
 
37
  SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS = {k for k,v in MODEL_META["model_meta"].items() if v.get("is_sentence_transformers_compatible", False)}
38
  MODELS_TO_SKIP = MODEL_META["models_to_skip"]
 
 
39
 
40
  PROPRIETARY_MODELS = {
41
  make_clickable_model(model, link=EXTERNAL_MODEL_TO_LINK.get(model, f"https://huggingface.co/spaces/{REPO_ID}"))
42
  for model in PROPRIETARY_MODELS
43
  }
44
-
45
  SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS = {
46
  make_clickable_model(model, link=EXTERNAL_MODEL_TO_LINK.get(model, f"https://huggingface.co/spaces/{REPO_ID}"))
47
  for model in SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS
48
  }
 
 
 
 
 
 
 
 
 
49
 
50
  TASK_TO_TASK_TYPE = {task_category: [] for task_category in TASKS}
51
  for board_config in BOARDS_CONFIG.values():
@@ -164,7 +181,13 @@ def get_mteb_data(tasks=["Clustering"], langs=[], datasets=[], fillna=True, add_
164
  # Initialize list to models that we cannot fetch metadata from
165
  df_list = []
166
  for model in EXTERNAL_MODEL_RESULTS:
167
- results_list = [res for task in tasks for res in EXTERNAL_MODEL_RESULTS[model][task][task_to_metric[task]]]
 
 
 
 
 
 
168
  if len(datasets) > 0:
169
  res = {k: v for d in results_list for k, v in d.items() if (k == "Model") or any([x in k for x in datasets])}
170
  elif langs:
@@ -383,7 +406,10 @@ for task in TASKS:
383
  data[task] = {"metric": TASKS_CONFIG[task]["metric_description"], "data": []}
384
 
385
  for board, board_config in BOARDS_CONFIG.items():
386
- board_pretty_name = f"{board_config['title']} leaderboard"
 
 
 
387
  acronym = board_config.get("acronym", None)
388
  board_icon = board_config.get("icon", None)
389
  if board_icon is None:
@@ -439,7 +465,7 @@ function(goalUrlObject) {
439
  def update_url_task(event: gr.SelectData, current_task_language: dict, language_per_task: dict):
440
  current_task_language["task"] = event.target.id
441
  # Either use the cached language for this task or the 1st language
442
- current_task_language["language"] = language_per_task.get(event.target.id, event.target.children[0].children[0].id)
443
  return current_task_language, language_per_task
444
 
445
  def update_url_language(event: gr.SelectData, current_task_language: dict, language_per_task: dict):
@@ -461,6 +487,8 @@ MODEL_TYPES = [
461
  "Open",
462
  "Proprietary",
463
  "Sentence Transformers",
 
 
464
  ]
465
 
466
  def filter_data(search_query, model_types, model_sizes, *full_dataframes):
@@ -484,6 +512,10 @@ def filter_data(search_query, model_types, model_sizes, *full_dataframes):
484
  masks.append(df["Model"].isin(PROPRIETARY_MODELS))
485
  elif model_type == "Sentence Transformers":
486
  masks.append(df["Model"].isin(SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS))
 
 
 
 
487
  if masks:
488
  df = df[reduce(lambda a, b: a | b, masks)]
489
  else:
@@ -535,16 +567,16 @@ with gr.Blocks(css=css) as block:
535
  with gr.Tabs() as outer_tabs:
536
  # Store the tabs for updating them on load based on URL parameters
537
  tabs.append(outer_tabs)
538
-
539
  for task, task_values in data.items():
540
  metric = task_values["metric"]
541
  task_tab_id = task.lower().replace(" ", "-")
542
 
543
  # Overall, Bitext Mining, Classification, etc.
544
- with gr.Tab(task, id=task_tab_id) as task_tab:
 
545
  # For updating the 'task' in the URL
546
  task_tab.select(update_url_task, [current_task_language, language_per_task], [current_task_language, language_per_task]).then(None, [current_task_language], [], js=set_window_url_params)
547
-
548
  with gr.Tabs() as task_tabs:
549
  # Store the task tabs for updating them on load based on URL parameters
550
  tabs.append(task_tabs)
 
17
  BOARDS_CONFIG = LEADERBOARD_CONFIG["boards"]
18
 
19
  TASKS = list(TASKS_CONFIG.keys())
20
+ PRETTY_NAMES = {
21
+ "InstructionRetrieval": "Retrieval w/Instructions",
22
+ "PairClassification": "Pair Classification",
23
+ "BitextMining": "Bitext Mining",
24
+ }
25
 
26
  TASK_TO_METRIC = {k:v["metric"] for k,v in TASKS_CONFIG.items()}
27
 
 
39
  EXTERNAL_MODEL_TO_SEQLEN = {k: v["seq_len"] for k,v in MODEL_META["model_meta"].items() if v.get("seq_len", False)}
40
  EXTERNAL_MODEL_TO_SIZE = {k: v["size"] for k,v in MODEL_META["model_meta"].items() if v.get("size", False)}
41
  PROPRIETARY_MODELS = {k for k,v in MODEL_META["model_meta"].items() if v.get("is_proprietary", False)}
42
+ TASK_DESCRIPTIONS = {k: v["task_description"] for k,v in TASKS_CONFIG.items()}
43
+ TASK_DESCRIPTIONS["Overall"] = "Overall performance across MTEB tasks."
44
  SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS = {k for k,v in MODEL_META["model_meta"].items() if v.get("is_sentence_transformers_compatible", False)}
45
  MODELS_TO_SKIP = MODEL_META["models_to_skip"]
46
+ CROSS_ENCODERS = MODEL_META["cross_encoders"]
47
+ BI_ENCODERS = [k for k, _ in MODEL_META["model_meta"].items() if k not in CROSS_ENCODERS + ["bm25"]]
48
 
49
  PROPRIETARY_MODELS = {
50
  make_clickable_model(model, link=EXTERNAL_MODEL_TO_LINK.get(model, f"https://huggingface.co/spaces/{REPO_ID}"))
51
  for model in PROPRIETARY_MODELS
52
  }
 
53
  SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS = {
54
  make_clickable_model(model, link=EXTERNAL_MODEL_TO_LINK.get(model, f"https://huggingface.co/spaces/{REPO_ID}"))
55
  for model in SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS
56
  }
57
+ CROSS_ENCODERS = {
58
+ make_clickable_model(model, link=EXTERNAL_MODEL_TO_LINK.get(model, f"https://huggingface.co/spaces/{REPO_ID}"))
59
+ for model in CROSS_ENCODERS
60
+ }
61
+ BI_ENCODERS = {
62
+ make_clickable_model(model, link=EXTERNAL_MODEL_TO_LINK.get(model, f"https://huggingface.co/spaces/{REPO_ID}"))
63
+ for model in BI_ENCODERS
64
+ }
65
+
66
 
67
  TASK_TO_TASK_TYPE = {task_category: [] for task_category in TASKS}
68
  for board_config in BOARDS_CONFIG.values():
 
181
  # Initialize list to models that we cannot fetch metadata from
182
  df_list = []
183
  for model in EXTERNAL_MODEL_RESULTS:
184
+ results_list = []
185
+ for task in tasks:
186
+ # Not all models have InstructionRetrieval, other new tasks
187
+ if task not in EXTERNAL_MODEL_RESULTS[model]:
188
+ continue
189
+ results_list += EXTERNAL_MODEL_RESULTS[model][task][task_to_metric[task]]
190
+
191
  if len(datasets) > 0:
192
  res = {k: v for d in results_list for k, v in d.items() if (k == "Model") or any([x in k for x in datasets])}
193
  elif langs:
 
406
  data[task] = {"metric": TASKS_CONFIG[task]["metric_description"], "data": []}
407
 
408
  for board, board_config in BOARDS_CONFIG.items():
409
+ init_name = board_config["title"]
410
+ if init_name in PRETTY_NAMES:
411
+ init_name = PRETTY_NAMES[init_name]
412
+ board_pretty_name = f"{init_name} leaderboard"
413
  acronym = board_config.get("acronym", None)
414
  board_icon = board_config.get("icon", None)
415
  if board_icon is None:
 
465
  def update_url_task(event: gr.SelectData, current_task_language: dict, language_per_task: dict):
466
  current_task_language["task"] = event.target.id
467
  # Either use the cached language for this task or the 1st language
468
+ current_task_language["language"] = language_per_task.get(event.target.id, event.target.children[1].children[0].id)
469
  return current_task_language, language_per_task
470
 
471
  def update_url_language(event: gr.SelectData, current_task_language: dict, language_per_task: dict):
 
487
  "Open",
488
  "Proprietary",
489
  "Sentence Transformers",
490
+ "Cross-Encoders",
491
+ "Bi-Encoders"
492
  ]
493
 
494
  def filter_data(search_query, model_types, model_sizes, *full_dataframes):
 
512
  masks.append(df["Model"].isin(PROPRIETARY_MODELS))
513
  elif model_type == "Sentence Transformers":
514
  masks.append(df["Model"].isin(SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS))
515
+ elif model_type == "Cross-Encoders":
516
+ masks.append(df["Model"].isin(CROSS_ENCODERS))
517
+ elif model_type == "Bi-Encoders":
518
+ masks.append(df["Model"].isin(BI_ENCODERS))
519
  if masks:
520
  df = df[reduce(lambda a, b: a | b, masks)]
521
  else:
 
567
  with gr.Tabs() as outer_tabs:
568
  # Store the tabs for updating them on load based on URL parameters
569
  tabs.append(outer_tabs)
 
570
  for task, task_values in data.items():
571
  metric = task_values["metric"]
572
  task_tab_id = task.lower().replace(" ", "-")
573
 
574
  # Overall, Bitext Mining, Classification, etc.
575
+ pretty_task_name = task if task not in PRETTY_NAMES.keys() else PRETTY_NAMES[task]
576
+ with gr.Tab(pretty_task_name, id=task_tab_id) as task_tab:
577
  # For updating the 'task' in the URL
578
  task_tab.select(update_url_task, [current_task_language, language_per_task], [current_task_language, language_per_task]).then(None, [current_task_language], [], js=set_window_url_params)
579
+ gr.Markdown(TASK_DESCRIPTIONS[task])
580
  with gr.Tabs() as task_tabs:
581
  # Store the task tabs for updating them on load based on URL parameters
582
  tabs.append(task_tabs)
config.yaml CHANGED
@@ -7,34 +7,47 @@ tasks:
7
  icon: "🎌"
8
  metric: f1
9
  metric_description: "[F1](https://huggingface.co/spaces/evaluate-metric/f1)"
 
10
  Classification:
11
  icon: "❤️"
12
  metric: accuracy
13
  metric_description: "[Accuracy](https://huggingface.co/spaces/evaluate-metric/accuracy)"
 
14
  Clustering:
15
  icon: "✨"
16
  metric: v_measure
17
  metric_description: "Validity Measure (v_measure)"
 
18
  PairClassification:
19
  icon: "🎭"
20
  metric: cos_sim_ap
21
  metric_description: "Average Precision based on Cosine Similarities (cos_sim_ap)"
 
22
  Reranking:
23
  icon: "🥈"
24
  metric: map
25
  metric_description: "Mean Average Precision (MAP)"
 
26
  Retrieval:
27
  icon: "🔎"
28
  metric: ndcg_at_10
29
  metric_description: "Normalized Discounted Cumulative Gain @ k (ndcg_at_10)"
 
30
  STS:
31
  icon: "🤖"
32
  metric: cos_sim_spearman
33
  metric_description: "Spearman correlation based on cosine similarity"
 
34
  Summarization:
35
  icon: "📜"
36
  metric: cos_sim_spearman
37
  metric_description: "Spearman correlation based on cosine similarity"
 
 
 
 
 
 
38
  boards:
39
  en:
40
  title: English
@@ -250,6 +263,18 @@ boards:
250
  - MassiveIntentClassification (nb)
251
  - MassiveScenarioClassification (nb)
252
  - ScalaNbClassification
 
 
 
 
 
 
 
 
 
 
 
 
253
  law:
254
  title: Law
255
  language_long: "English, German, Chinese"
 
7
  icon: "🎌"
8
  metric: f1
9
  metric_description: "[F1](https://huggingface.co/spaces/evaluate-metric/f1)"
10
+ task_description: "Bitext mining is the task of finding parallel sentences in two languages."
11
  Classification:
12
  icon: "❤️"
13
  metric: accuracy
14
  metric_description: "[Accuracy](https://huggingface.co/spaces/evaluate-metric/accuracy)"
15
+ task_description: "Classification is the task of assigning a label to a text."
16
  Clustering:
17
  icon: "✨"
18
  metric: v_measure
19
  metric_description: "Validity Measure (v_measure)"
20
+ task_description: "Clustering is the task of grouping similar documents together."
21
  PairClassification:
22
  icon: "🎭"
23
  metric: cos_sim_ap
24
  metric_description: "Average Precision based on Cosine Similarities (cos_sim_ap)"
25
+ task_description: "Pair classification is the task of determining whether two texts are similar."
26
  Reranking:
27
  icon: "🥈"
28
  metric: map
29
  metric_description: "Mean Average Precision (MAP)"
30
+ task_description: "Reranking is the task of reordering a list of documents to improve relevance."
31
  Retrieval:
32
  icon: "🔎"
33
  metric: ndcg_at_10
34
  metric_description: "Normalized Discounted Cumulative Gain @ k (ndcg_at_10)"
35
+ task_description: "Retrieval is the task of finding relevant documents for a query."
36
  STS:
37
  icon: "🤖"
38
  metric: cos_sim_spearman
39
  metric_description: "Spearman correlation based on cosine similarity"
40
+ task_description: "Semantic Textual Similarity is the task of determining how similar two texts are."
41
  Summarization:
42
  icon: "📜"
43
  metric: cos_sim_spearman
44
  metric_description: "Spearman correlation based on cosine similarity"
45
+ task_description: "Summarization is the task of generating a summary of a text."
46
+ InstructionRetrieval:
47
+ icon: "🔎📋"
48
+ metric: "p-MRR"
49
+ metric_description: "paired mean reciprocal rank"
50
+ task_description: "Retrieval w/Instructions is the task of finding relevant documents for a query that has detailed instructions."
51
  boards:
52
  en:
53
  title: English
 
263
  - MassiveIntentClassification (nb)
264
  - MassiveScenarioClassification (nb)
265
  - ScalaNbClassification
266
+ instructions:
267
+ title: English
268
+ language_long: "English"
269
+ has_overall: false
270
+ acronym: null
271
+ icon: null
272
+ credits: "[Orion Weller, FollowIR](https://arxiv.org/abs/2403.15246)"
273
+ tasks:
274
+ InstructionRetrieval:
275
+ - Robust04InstructionRetrieval
276
+ - News21InstructionRetrieval
277
+ - Core17InstructionRetrieval
278
  law:
279
  title: Law
280
  language_long: "English, German, Chinese"
model_meta.yaml CHANGED
@@ -47,6 +47,20 @@ model_meta:
47
  is_external: true
48
  is_proprietary: false
49
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
 
 
 
 
 
 
 
 
50
  LASER2:
51
  link: https://github.com/facebookresearch/LASER
52
  seq_len: N/A
@@ -263,6 +277,12 @@ model_meta:
263
  is_external: true
264
  is_proprietary: false
265
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
266
  camembert-base:
267
  link: https://huggingface.co/almanach/camembert-base
268
  seq_len: 512
@@ -359,6 +379,14 @@ model_meta:
359
  is_external: true
360
  is_proprietary: false
361
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
 
 
362
  e5-base:
363
  link: https://huggingface.co/intfloat/e5-base
364
  seq_len: 512
@@ -367,6 +395,14 @@ model_meta:
367
  is_external: true
368
  is_proprietary: false
369
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
 
 
370
  e5-large:
371
  link: https://huggingface.co/intfloat/e5-large
372
  seq_len: 512
@@ -407,6 +443,22 @@ model_meta:
407
  is_external: true
408
  is_proprietary: false
409
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
410
  flaubert_base_cased:
411
  link: https://huggingface.co/flaubert/flaubert_base_cased
412
  seq_len: 512
@@ -535,6 +587,22 @@ model_meta:
535
  is_external: true
536
  is_proprietary: false
537
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
538
  komninos:
539
  link: https://huggingface.co/sentence-transformers/average_word_embeddings_komninos
540
  seq_len: N/A
@@ -543,6 +611,14 @@ model_meta:
543
  is_external: true
544
  is_proprietary: false
545
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
 
 
546
  luotuo-bert-medium:
547
  link: https://huggingface.co/silk-road/luotuo-bert-medium
548
  seq_len: 512
@@ -567,6 +643,14 @@ model_meta:
567
  is_external: true
568
  is_proprietary: false
569
  is_sentence_transformers_compatible: true
 
 
 
 
 
 
 
 
570
  mistral-embed:
571
  link: https://docs.mistral.ai/guides/embeddings
572
  seq_len: null
@@ -575,6 +659,30 @@ model_meta:
575
  is_external: true
576
  is_proprietary: true
577
  is_sentence_transformers_compatible: false
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
578
  msmarco-bert-co-condensor:
579
  link: https://huggingface.co/sentence-transformers/msmarco-bert-co-condensor
580
  seq_len: 512
@@ -903,6 +1011,22 @@ model_meta:
903
  is_external: true
904
  is_proprietary: true
905
  is_sentence_transformers_compatible: false
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
906
  text2vec-base-chinese:
907
  link: https://huggingface.co/shibing624/text2vec-base-chinese
908
  seq_len: 512
@@ -1184,3 +1308,13 @@ models_to_skip:
1184
  - michaelfeil/ct2fast-gte-large
1185
  - gizmo-ai/Cohere-embed-multilingual-v3.0
1186
  - McGill-NLP/LLM2Vec-Llama-2-7b-chat-hf-mntp-unsup-simcse
 
 
 
 
 
 
 
 
 
 
 
47
  is_external: true
48
  is_proprietary: false
49
  is_sentence_transformers_compatible: true
50
+ FollowIR-7B:
51
+ link: https://huggingface.co/jhu-clsp/FollowIR-7B
52
+ seq_len: 4096
53
+ size: 7240
54
+ is_external: true
55
+ is_propietary: false
56
+ is_sentence_transformer_compatible: false
57
+ GritLM-7B:
58
+ link: https://huggingface.co/GritLM/GritLM-7B
59
+ seq_len: 4096
60
+ size: 7240
61
+ is_external: true
62
+ is_propietary: false
63
+ is_sentence_transformer_compatible: false
64
  LASER2:
65
  link: https://github.com/facebookresearch/LASER
66
  seq_len: N/A
 
277
  is_external: true
278
  is_proprietary: false
279
  is_sentence_transformers_compatible: true
280
+ bm25:
281
+ link: https://en.wikipedia.org/wiki/Okapi_BM25
282
+ size: 0
283
+ is_external: true
284
+ is_proprietary: false
285
+ is_sentence_transformers_compatible: false
286
  camembert-base:
287
  link: https://huggingface.co/almanach/camembert-base
288
  seq_len: 512
 
379
  is_external: true
380
  is_proprietary: false
381
  is_sentence_transformers_compatible: true
382
+ e5-base-v2:
383
+ link: https://huggingface.co/intfloat/e5-base-v2
384
+ seq_len: 512
385
+ size: 110
386
+ dim: 768
387
+ is_external: true
388
+ is_proprietary: false
389
+ is_sentence_transformers_compatible: true
390
  e5-base:
391
  link: https://huggingface.co/intfloat/e5-base
392
  seq_len: 512
 
395
  is_external: true
396
  is_proprietary: false
397
  is_sentence_transformers_compatible: true
398
+ e5-large-v2:
399
+ link: https://huggingface.co/intfloat/e5-large-v2
400
+ seq_len: 512
401
+ size: 335
402
+ dim: 1024
403
+ is_external: true
404
+ is_proprietary: false
405
+ is_sentence_transformers_compatible: true
406
  e5-large:
407
  link: https://huggingface.co/intfloat/e5-large
408
  seq_len: 512
 
443
  is_external: true
444
  is_proprietary: false
445
  is_sentence_transformers_compatible: true
446
+ flan-t5-base:
447
+ link: https://huggingface.co/google/flan-t5-base
448
+ seq_len: 512
449
+ size: 220
450
+ dim: -1
451
+ is_external: true
452
+ is_proprietary: false
453
+ is_sentence_transformers_compatible: true
454
+ flan-t5-large:
455
+ link: https://huggingface.co/google/flan-t5-large
456
+ seq_len: 512
457
+ size: 770
458
+ dim: -1
459
+ is_external: true
460
+ is_proprietary: false
461
+ is_sentence_transformers_compatible: true
462
  flaubert_base_cased:
463
  link: https://huggingface.co/flaubert/flaubert_base_cased
464
  seq_len: 512
 
587
  is_external: true
588
  is_proprietary: false
589
  is_sentence_transformers_compatible: true
590
+ instructor-base:
591
+ link: https://huggingface.co/hkunlp/instructor-base
592
+ seq_len: N/A
593
+ size: 110
594
+ dim: 768
595
+ is_external: true
596
+ is_proprietary: false
597
+ is_sentence_transformers_compatible: true
598
+ instructor-xl:
599
+ link: https://huggingface.co/hkunlp/instructor-xl
600
+ seq_len: N/A
601
+ size: 1241
602
+ dim: 768
603
+ is_external: true
604
+ is_proprietary: false
605
+ is_sentence_transformers_compatible: true
606
  komninos:
607
  link: https://huggingface.co/sentence-transformers/average_word_embeddings_komninos
608
  seq_len: N/A
 
611
  is_external: true
612
  is_proprietary: false
613
  is_sentence_transformers_compatible: true
614
+ llama-2-7b-chat:
615
+ link: https://huggingface.co/meta-llama/Llama-2-7b-chat-hf
616
+ seq_len: 4096
617
+ size: 7000
618
+ dim: -1
619
+ is_external: true
620
+ is_proprietary: false
621
+ is_sentence_transformers_compatible: false
622
  luotuo-bert-medium:
623
  link: https://huggingface.co/silk-road/luotuo-bert-medium
624
  seq_len: 512
 
643
  is_external: true
644
  is_proprietary: false
645
  is_sentence_transformers_compatible: true
646
+ mistral-7b-instruct-v0.2:
647
+ link: https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2
648
+ seq_len: 4096
649
+ size: 7240
650
+ dim: -1
651
+ is_external: true
652
+ is_proprietary: false
653
+ is_sentence_transformers_compatible: false
654
  mistral-embed:
655
  link: https://docs.mistral.ai/guides/embeddings
656
  seq_len: null
 
659
  is_external: true
660
  is_proprietary: true
661
  is_sentence_transformers_compatible: false
662
+ monobert-large-msmarco:
663
+ link: https://huggingface.co/castorini/monobert-large-msmarco
664
+ seq_len: 512
665
+ size: 770
666
+ dim: -1
667
+ is_external: true
668
+ is_proprietary: false
669
+ is_sentence_transformers_compatible: false
670
+ monot5-3b-msmarco-10k:
671
+ link: https://huggingface.co/castorini/monot5-3b-msmarco-10k
672
+ seq_len: 512
673
+ size: 2480
674
+ dim: -1
675
+ is_external: true
676
+ is_proprietary: false
677
+ is_sentence_transformers_compatible: false
678
+ monot5-base-msmarco-10k:
679
+ link: https://huggingface.co/castorini/monot5-base-msmarco-10k
680
+ seq_len: 512
681
+ size: 220
682
+ dim: -1
683
+ is_external: true
684
+ is_proprietary: false
685
+ is_sentence_transformers_compatible: false
686
  msmarco-bert-co-condensor:
687
  link: https://huggingface.co/sentence-transformers/msmarco-bert-co-condensor
688
  seq_len: 512
 
1011
  is_external: true
1012
  is_proprietary: true
1013
  is_sentence_transformers_compatible: false
1014
+ tart-dual-contriever-msmarco:
1015
+ link: https://huggingface.co/orionweller/tart-dual-contriever-msmarco
1016
+ seq_len: 512
1017
+ size: 110
1018
+ dim: 768
1019
+ is_external: true
1020
+ is_proprietary: false
1021
+ is_sentence_transformers_compatible: false
1022
+ tart-full-flan-t5-xl:
1023
+ link: https://huggingface.co/facebook/tart-full-flan-t5-xl
1024
+ seq_len: 512
1025
+ size: 2480
1026
+ dim: -1
1027
+ is_external: true
1028
+ is_proprietary: false
1029
+ is_sentence_transformers_compatible: false
1030
  text2vec-base-chinese:
1031
  link: https://huggingface.co/shibing624/text2vec-base-chinese
1032
  seq_len: 512
 
1308
  - michaelfeil/ct2fast-gte-large
1309
  - gizmo-ai/Cohere-embed-multilingual-v3.0
1310
  - McGill-NLP/LLM2Vec-Llama-2-7b-chat-hf-mntp-unsup-simcse
1311
+ cross_encoders:
1312
+ - FollowIR-7B
1313
+ - flan-t5-base
1314
+ - flan-t5-large
1315
+ - monobert-large-msmarco
1316
+ - monot5-3b-msmarco-10k
1317
+ - monot5-base-msmarco-10k
1318
+ - llama-2-7b-chat
1319
+ - mistral-7b-instruct-v0.2
1320
+ - tart-full-flan-t5-xl