Skip to content

[mieb] Fix Jina CLIP #1349

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
gowitheflow-1998 merged 1 commit into from
Oct 28, 2024
Merged

[mieb] Fix Jina CLIP #1349

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 9 additions & 35 deletions mteb/models/jina_clip.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,8 +11,6 @@

from mteb.model_meta import ModelMeta

# from mteb.models.text_formatting_utils import corpus_to_texts


class JinaCLIPModelWrapper:
def __init__(
Expand Down Expand Up @@ -48,9 +46,13 @@ def get_image_embeddings(

if isinstance(images, DataLoader):
with torch.no_grad():
import torchvision.transforms.functional as F

for batch in tqdm(images):
image_outputs = self.model.encode_image(
batch, convert_to_numpy=False, convert_to_tensor=True
[F.to_pil_image(b.to("cpu")) for b in batch],
convert_to_numpy=False,
convert_to_tensor=True,
)
all_image_embeddings.append(image_outputs.cpu())
else:
Expand Down Expand Up @@ -126,48 +128,20 @@ def encode( # type: ignore
batch_size: int = 32,
**kwargs: Any,
):
if "prompt_name" in kwargs:
kwargs.pop("prompt_name")
if "task_name" in kwargs:
kwargs.pop("task_name")
return self.model.encode_text(sentences, batch_size=batch_size, **kwargs)

# def encode_queries(self, queries: list[str], batch_size: int = 32, **kwargs: Any):
# if "prompt_name" in kwargs:
# kwargs.pop("prompt_name")
# sentences = [
# "Represent this sentence for searching relevant passages: " + sentence
# for sentence in queries
# ]
# emb = self.encode(
# sentences, batch_size=batch_size, normalize_embeddings=True, **kwargs
# )
# return emb

# def encode_corpus(
# self,
# corpus: list[dict[str, str]] | dict[str, list[str]],
# batch_size: int = 32,
# **kwargs: Any,
# ):
# if "prompt_name" in kwargs:
# kwargs.pop("prompt_name")
# sentences = corpus_to_texts(corpus)
# emb = self.encode(
# sentences, batch_size=batch_size, normalize_embeddings=True, **kwargs
# )
# return emb


from mteb.model_meta import sentence_transformers_loader

jina_clip_v1 = ModelMeta(
loader=partial(
sentence_transformers_loader,
JinaCLIPModelWrapper,
model_name="jinaai/jina-clip-v1",
),
name="jinaai/jina-clip-v1",
languages=["eng_Latn"],
open_source=True,
revision="1cbe5e8b11ea3728df0b610d5453dfe739804aa9",
revision="06150c7c382d7a4faedc7d5a0d8cdb59308968f4",
release_date="2024-05-30",
)

Expand Down
186 changes: 186 additions & 0 deletions ...ieb/jinaai__jina-clip-v1/06150c7c382d7a4faedc7d5a0d8cdb59308968f4/MSCOCOI2TRetrieval.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,186 @@
{
"dataset_revision": "cca3a3e223763e6519a4d68936bc9279034d75d2",
"evaluation_time": 57.833680629730225,
"kg_co2_emissions": null,
"mteb_version": "1.14.21",
"scores": {
"test": [
{
"cv_recall_at_1": 0.544,
"cv_recall_at_10": 0.8592,
"cv_recall_at_100": 0.993,
"cv_recall_at_1000": 0.9998,
"cv_recall_at_20": 0.922,
"cv_recall_at_3": 0.7132,
"cv_recall_at_5": 0.7826,
"hf_subset": "default",
"languages": [
"eng-Latn"
],
"main_score": 0.48279,
"map_at_1": 0.10886,
"map_at_10": 0.36753,
"map_at_100": 0.41438,
"map_at_1000": 0.41751,
"map_at_20": 0.39363,
"map_at_3": 0.2442,
"map_at_5": 0.31641,
"mrr_at_1": 0.544,
"mrr_at_10": 0.6441334920634935,
"mrr_at_100": 0.6505229831896451,
"mrr_at_1000": 0.6505722508994467,
"mrr_at_20": 0.6485590829105544,
"mrr_at_3": 0.6178333333333357,
"mrr_at_5": 0.6337633333333345,
"nauc_cv_recall_at_1000_diff1": -1.1517273576107072,
"nauc_cv_recall_at_1000_max": 1.0,
"nauc_cv_recall_at_1000_std": 1.0,
"nauc_cv_recall_at_100_diff1": -0.001093770841666523,
"nauc_cv_recall_at_100_max": 0.4157129518474038,
"nauc_cv_recall_at_100_std": -0.3355875683606853,
"nauc_cv_recall_at_10_diff1": 0.3116322316328111,
"nauc_cv_recall_at_10_max": 0.22344380665652067,
"nauc_cv_recall_at_10_std": -0.5484061734930179,
"nauc_cv_recall_at_1_diff1": 0.3620563318972254,
"nauc_cv_recall_at_1_max": 0.22304702323816303,
"nauc_cv_recall_at_1_std": -0.39325847713016515,
"nauc_cv_recall_at_20_diff1": 0.3174411644999864,
"nauc_cv_recall_at_20_max": 0.26129067011419665,
"nauc_cv_recall_at_20_std": -0.5873327108621224,
"nauc_cv_recall_at_3_diff1": 0.3190785659299513,
"nauc_cv_recall_at_3_max": 0.22356865681802765,
"nauc_cv_recall_at_3_std": -0.45729779105093976,
"nauc_cv_recall_at_5_diff1": 0.32854938408837603,
"nauc_cv_recall_at_5_max": 0.21763614808463352,
"nauc_cv_recall_at_5_std": -0.4901185955745346,
"nauc_map_at_1000_diff1": 0.21096215655518416,
"nauc_map_at_1000_max": 0.26437483183618576,
"nauc_map_at_1000_std": -0.3656676562792446,
"nauc_map_at_100_diff1": 0.21094575808032542,
"nauc_map_at_100_max": 0.2647084321742531,
"nauc_map_at_100_std": -0.36659141600302986,
"nauc_map_at_10_diff1": 0.20614205326831164,
"nauc_map_at_10_max": 0.25564678162479904,
"nauc_map_at_10_std": -0.3625738173870285,
"nauc_map_at_1_diff1": 0.3621079914040161,
"nauc_map_at_1_max": 0.22314949459519212,
"nauc_map_at_1_std": -0.39289670930715714,
"nauc_map_at_20_diff1": 0.2088651049333622,
"nauc_map_at_20_max": 0.2606844896581087,
"nauc_map_at_20_std": -0.36749366718366866,
"nauc_map_at_3_diff1": 0.23125083203784524,
"nauc_map_at_3_max": 0.23835700021972306,
"nauc_map_at_3_std": -0.3525765196558998,
"nauc_map_at_5_diff1": 0.20395917414865206,
"nauc_map_at_5_max": 0.24890333320816935,
"nauc_map_at_5_std": -0.3551844385945933,
"nauc_mrr_at_1000_diff1": 0.3462109269235247,
"nauc_mrr_at_1000_max": 0.22341427843542805,
"nauc_mrr_at_1000_std": -0.421124236204143,
"nauc_mrr_at_100_diff1": 0.34615985793627224,
"nauc_mrr_at_100_max": 0.2234417011721278,
"nauc_mrr_at_100_std": -0.42112571190402276,
"nauc_mrr_at_10_diff1": 0.3457260635691663,
"nauc_mrr_at_10_max": 0.22268948650151105,
"nauc_mrr_at_10_std": -0.4235752064832113,
"nauc_mrr_at_1_diff1": 0.3620563318972254,
"nauc_mrr_at_1_max": 0.22304702323816303,
"nauc_mrr_at_1_std": -0.39325847713016515,
"nauc_mrr_at_20_diff1": 0.34612102767284886,
"nauc_mrr_at_20_max": 0.22330420489291178,
"nauc_mrr_at_20_std": -0.42260480382387744,
"nauc_mrr_at_3_diff1": 0.3436568869473764,
"nauc_mrr_at_3_max": 0.22268159334592896,
"nauc_mrr_at_3_std": -0.41793547872411135,
"nauc_mrr_at_5_diff1": 0.3462959273283977,
"nauc_mrr_at_5_max": 0.22190940766444717,
"nauc_mrr_at_5_std": -0.42173732217738524,
"nauc_ndcg_at_1000_diff1": 0.24740773356795673,
"nauc_ndcg_at_1000_max": 0.2633139225090684,
"nauc_ndcg_at_1000_std": -0.3825727328439968,
"nauc_ndcg_at_100_diff1": 0.2439247459069718,
"nauc_ndcg_at_100_max": 0.27221683198323254,
"nauc_ndcg_at_100_std": -0.38503378892448387,
"nauc_ndcg_at_10_diff1": 0.2322351990974726,
"nauc_ndcg_at_10_max": 0.25133559217064416,
"nauc_ndcg_at_10_std": -0.38906206049843284,
"nauc_ndcg_at_1_diff1": 0.3620563318972254,
"nauc_ndcg_at_1_max": 0.22304702323816303,
"nauc_ndcg_at_1_std": -0.39325847713016515,
"nauc_ndcg_at_20_diff1": 0.2368585564555519,
"nauc_ndcg_at_20_max": 0.2586236191288364,
"nauc_ndcg_at_20_std": -0.39571963457704934,
"nauc_ndcg_at_3_diff1": 0.25241282801383136,
"nauc_ndcg_at_3_max": 0.23505398892057644,
"nauc_ndcg_at_3_std": -0.37119292314398367,
"nauc_ndcg_at_5_diff1": 0.2325478110853486,
"nauc_ndcg_at_5_max": 0.24393791588243935,
"nauc_ndcg_at_5_std": -0.37684221871528933,
"nauc_precision_at_1000_diff1": -0.060776645601850005,
"nauc_precision_at_1000_max": 0.45061728395062717,
"nauc_precision_at_1000_std": 0.49837165139806966,
"nauc_precision_at_100_diff1": 0.16399269478318815,
"nauc_precision_at_100_max": 0.3174441756422564,
"nauc_precision_at_100_std": -0.2719045017872643,
"nauc_precision_at_10_diff1": 0.1834674594026287,
"nauc_precision_at_10_max": 0.2475111637901415,
"nauc_precision_at_10_std": -0.36964525963118233,
"nauc_precision_at_1_diff1": 0.3620563318972254,
"nauc_precision_at_1_max": 0.22304702323816303,
"nauc_precision_at_1_std": -0.39325847713016515,
"nauc_precision_at_20_diff1": 0.18190648500233034,
"nauc_precision_at_20_max": 0.2543371579459487,
"nauc_precision_at_20_std": -0.3684848720360581,
"nauc_precision_at_3_diff1": 0.2216308838348382,
"nauc_precision_at_3_max": 0.23889515868420422,
"nauc_precision_at_3_std": -0.3624204875957743,
"nauc_precision_at_5_diff1": 0.19290462095741936,
"nauc_precision_at_5_max": 0.2465663979201694,
"nauc_precision_at_5_std": -0.36446145380745226,
"nauc_recall_at_1000_diff1": -0.04813353912994217,
"nauc_recall_at_1000_max": 0.47849854227402444,
"nauc_recall_at_1000_std": 0.5228805808037352,
"nauc_recall_at_100_diff1": 0.16501981261932044,
"nauc_recall_at_100_max": 0.3183331361282548,
"nauc_recall_at_100_std": -0.2726147056630754,
"nauc_recall_at_10_diff1": 0.1835675928057196,
"nauc_recall_at_10_max": 0.24766746744342197,
"nauc_recall_at_10_std": -0.3696577837871595,
"nauc_recall_at_1_diff1": 0.3621079914040161,
"nauc_recall_at_1_max": 0.22314949459519212,
"nauc_recall_at_1_std": -0.39289670930715714,
"nauc_recall_at_20_diff1": 0.18207077982715833,
"nauc_recall_at_20_max": 0.254625446742885,
"nauc_recall_at_20_std": -0.36840440513586076,
"nauc_recall_at_3_diff1": 0.2218119797611259,
"nauc_recall_at_3_max": 0.23903410739410114,
"nauc_recall_at_3_std": -0.3625477824595922,
"nauc_recall_at_5_diff1": 0.19309140539859648,
"nauc_recall_at_5_max": 0.2467762588423194,
"nauc_recall_at_5_std": -0.36457390258550315,
"ndcg_at_1": 0.544,
"ndcg_at_10": 0.48279,
"ndcg_at_100": 0.6064,
"ndcg_at_1000": 0.63735,
"ndcg_at_20": 0.53283,
"ndcg_at_3": 0.47129,
"ndcg_at_5": 0.41454,
"precision_at_1": 0.544,
"precision_at_10": 0.24992,
"precision_at_100": 0.04259,
"precision_at_1000": 0.00496,
"precision_at_20": 0.15413,
"precision_at_3": 0.45027,
"precision_at_5": 0.37584,
"recall_at_1": 0.10886,
"recall_at_10": 0.50001,
"recall_at_100": 0.85215,
"recall_at_1000": 0.99216,
"recall_at_20": 0.6168,
"recall_at_3": 0.27025,
"recall_at_5": 0.37596
}
]
},
"task_name": "MSCOCOI2TRetrieval"
}
26 changes: 26 additions & 0 deletions results-mieb/jinaai__jina-clip-v1/06150c7c382d7a4faedc7d5a0d8cdb59308968f4/STS14.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
{
"dataset_revision": "6031580fec1f6af667f0bd2da0a551cf4f0b2375",
"evaluation_time": 4.692688703536987,
"kg_co2_emissions": null,
"mteb_version": "1.14.21",
"scores": {
"test": [
{
"cosine_pearson": 0.8302122731320196,
"cosine_spearman": 0.7867521693830515,
"euclidean_pearson": 0.8140308674026765,
"euclidean_spearman": 0.7867522096109674,
"hf_subset": "default",
"languages": [
"eng-Latn"
],
"main_score": 0.7867521693830515,
"manhattan_pearson": 0.813071258051059,
"manhattan_spearman": 0.7856621071659424,
"pearson": 0.8302122731320196,
"spearman": 0.7867521693830515
}
]
},
"task_name": "STS14"
}
1 change: 1 addition & 0 deletions results-mieb/jinaai__jina-clip-v1/06150c7c382d7a4faedc7d5a0d8cdb59308968f4/model_meta.json
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"name": "jinaai/jina-clip-v1", "revision": "06150c7c382d7a4faedc7d5a0d8cdb59308968f4", "release_date": "2024-05-30", "languages": ["eng_Latn"], "n_parameters": null, "memory_usage": null, "max_tokens": null, "embed_dim": null, "license": null, "open_weights": null, "public_training_data": null, "public_training_code": null, "framework": [], "reference": null, "similarity_fn_name": null, "use_instuctions": null, "zero_shot_benchmarks": null, "loader": "JinaCLIPModelWrapper"}