Add new models nvidia, gte, linq

[AlexeyVatolin]

Added two gte, nvidia and linq models to model registry

Checklist

• Run tests locally to make sure nothing is broken using make test.
• Run the formatter to format the code using make lint.

Adding a model checklist

• I have filled out the ModelMeta object to the extent possible
• I have ensured that my model can be loaded using
  • mteb.get_model(model_name, revision) and
  • mteb.get_model_meta(model_name, revision)
• I have tested the implementation works on a representative set of tasks.

For testing I took the examples from the huggingface model card and compared the results with the results from the mteb model registry. In all new models, the text distance scores matches in at least 3 figures

[Samoed]

Great changes! Can you submit results?

[AlexeyVatolin]

Great changes! Can you submit results?

What do you mean, results on one arbitrary task from English MTEB?

[Samoed]

On some tasks from leaderboard to make sure that implementation matching

[AlexeyVatolin]

@Samoed, I computed scores on tasks as in your previous pull request with the addition of models (#1319)

Classification

	source	AmazonCounterfactualClassification	EmotionClassification	ToxicConversationsClassification
Linq-Embed-Mistral	Leaderboard	84.43	51.82	71.29
Linq-Embed-Mistral	Pull request	84.94	56.45	71.82
NV-Embed-v1	Leaderboard	95.12	91.7	92.6
NV-Embed-v1	Pull request	71.03	79.26	78.96
NV-Embed-v2	Leaderboard	94.28	93.38	92.74
NV-Embed-v2	Pull request	79.28	64.79	76.3
gte-Qwen1.5-7B-instruct	Leaderboard	83.16	54.53	78.75
gte-Qwen1.5-7B-instruct	Pull request	81.78	54.91	77.25
gte-Qwen2-1.5B-instruct	Leaderboard	83.99	61.37	82.66
gte-Qwen2-1.5B-instruct	Pull request	82.42	65.66	84.54

Clustering

	source	ArxivClusteringS2S	RedditClustering
Linq-Embed-Mistral	Leaderboard	47.3	61.52
Linq-Embed-Mistral	Pull request	47.61	60.94
NV-Embed-v1	Leaderboard	49.59	63.2
NV-Embed-v1	Pull request	48.31	52.29
NV-Embed-v2	Leaderboard	51.26	71.1
NV-Embed-v2	Pull request	46.98	55.58
gte-Qwen1.5-7B-instruct	Leaderboard	51.45	73.37
gte-Qwen1.5-7B-instruct	Pull request	53.57	80.12
gte-Qwen2-1.5B-instruct	Leaderboard	45.01	55.82
gte-Qwen2-1.5B-instruct	Pull request	44.61	51.36

PairClassification

	source	SprintDuplicateQuestions	TwitterSemEval2015
Linq-Embed-Mistral	Leaderboard	96.11	81.52
Linq-Embed-Mistral	Pull request	94.66	77.09
NV-Embed-v1	Leaderboard	95.94	79
NV-Embed-v1	Pull request	95.93	71.6
NV-Embed-v2	Leaderboard	97.02	81.11
NV-Embed-v2	Pull request	96.99	73.33
gte-Qwen1.5-7B-instruct	Leaderboard	96.07	79.36
gte-Qwen1.5-7B-instruct	Pull request	20.53	37.15
gte-Qwen2-1.5B-instruct	Leaderboard	95.32	79.64
gte-Qwen2-1.5B-instruct	Pull request	29.5	42.26

Reranking

	source	SciDocsRR	AskUbuntuDupQuestions
Linq-Embed-Mistral	Leaderboard	86.4	66.82
Linq-Embed-Mistral	Pull request	84.52	62.36
NV-Embed-v1	Leaderboard	87.26	67.5
NV-Embed-v1	Pull request	86.29	65.27
NV-Embed-v2	Leaderboard	87.59	67.46
NV-Embed-v2	Pull request	85.45	64.94
gte-Qwen1.5-7B-instruct	Leaderboard	87.89	66
gte-Qwen1.5-7B-instruct	Pull request	57.67	45.19
gte-Qwen2-1.5B-instruct	Leaderboard	86.52	64.55
gte-Qwen2-1.5B-instruct	Pull request	67.05	48.91

Retrieval

	source	SCIDOCS	SciFact
Linq-Embed-Mistral	Leaderboard	21.93	78.32
Linq-Embed-Mistral	Pull request	22.08	78.32
NV-Embed-v1	Leaderboard	20.19	78.43
NV-Embed-v1	Pull request	20.07	78.13
NV-Embed-v2	Leaderboard	21.9	80.13
NV-Embed-v2	Pull request	21.67	80.11
gte-Qwen1.5-7B-instruct	Leaderboard	27.69	75.31
gte-Qwen1.5-7B-instruct	Pull request	26.34	75.8
gte-Qwen2-1.5B-instruct	Leaderboard	24.98	78.44
gte-Qwen2-1.5B-instruct	Pull request	23.41	77.47

STS

	source	STS16	STSBenchmark
Linq-Embed-Mistral	Leaderboard	87.37	88.81
Linq-Embed-Mistral	Pull request	87.25	88.66
NV-Embed-v1	Leaderboard	84.77	86.14
NV-Embed-v1	Pull request	78.2	80.25
NV-Embed-v2	Leaderboard	86.77	88.41
NV-Embed-v2	Pull request	82.79	83.56
gte-Qwen1.5-7B-instruct	Leaderboard	86.39	87.35
gte-Qwen1.5-7B-instruct	Pull request	85.98	86.86
gte-Qwen2-1.5B-instruct	Leaderboard	85.45	86.38
gte-Qwen2-1.5B-instruct	Pull request	84.71	84.71

Summarization

	source	SummEval
Linq-Embed-Mistral	Leaderboard	30.98
Linq-Embed-Mistral	Pull request	30.39
NV-Embed-v1	Leaderboard	31.2
NV-Embed-v1	Pull request	29.37
NV-Embed-v2	Leaderboard	30.7
NV-Embed-v2	Pull request	30.42
gte-Qwen1.5-7B-instruct	Leaderboard	31.46
gte-Qwen1.5-7B-instruct	Pull request	31.22
gte-Qwen2-1.5B-instruct	Leaderboard	31.17
gte-Qwen2-1.5B-instruct	Pull request	30.5

I see a big difference in the results for the gte-Qwen models. Perhaps this is related to the prompts, gte uses prompt only for query, but in MTEB prompt is used for both query and document. Of the models I'm adding, the Linq-Embed-Mistral model has the least difference in scores. I think it can be merged without any changes. For other models, I think it is worth checking what the results are when you use the prompt only for query. @Samoed, what do you think?

[Samoed]

Great! The results on classification for NV-Embed show a significant gap as well. I think a wrapper can be created for gte-Qwen to add instructions only to the query. However, it's a bit strange that prompts seem to make performance worse on PairClassification.