nomic_embed_fine_tune_law

SentenceTransformer based on nomic-ai/nomic-embed-text-v1.5

This is a sentence-transformers model finetuned from nomic-ai/nomic-embed-text-v1.5. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: nomic-ai/nomic-embed-text-v1.5 <!-- at revision 679199c2575b5bfe93b06161d06cd7c16ebe4124 -->
• Maximum Sequence Length: 8192 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity

<!-- - **Training Dataset:** Unknown --> <!-- - **Language:** Unknown --> <!-- - **License:** Unknown -->

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 8192, 'do_lower_case': False}) with Transformer model: NomicBertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("simonosgoode/nomic_embed_fine_tune_law")
# Run inference
sentences = [
    'cluster: CONCLUSION:  Bacon St-Onge v. Conseil des Innus de Pessamit\nCourt (s) Database\nFederal Court Decisions\nDate\n2018-06-22\nNeutral citation\n2018 FC 655\nFile numbers\nT-2135-16\nDecision Content\nDate: 20180622\nDocket: T-2135-16\nCitation: 2018 FC 655\n[ENGLISH TRANSLATION]\nMontréal, Quebec, June 22, 2018\nPRESENT: The Honourable Madam Justice St-Louis\nBETWEEN:\nJÉRÔME BACON ST-ONGE\nApplicant\nand\nLE CONSEIL DES INNUS DE PESSAMIT\nRENÉ SIMON\nÉRIC CANAPÉ\nGÉRALD HERVIEUX\nDIANE RIVERIN\nJEAN-NOËL RIVERIN\nRAYMOND ROUSSELOT\nMARIELLE VACHON\nRespondents\nORDER AND REASONS\nI. Background\n[1] On December 21, 2017, the Court upheld the application for judicial review submitted by the Applicant, Jérôme Bacon St-Onge and, in particular, revoked the resolution adopted by the band council on March 8, 2016, adjudged the 2015 Code to be invalid, and voided the election held on August 17, 2016. The Court then asked the parties to make submissions concerning costs.\n[2] On January 22, 2018, the Respondents filed an appeal of this judgment with the Federal Court of Appeal [FCA]. At the same time, they also filed a motion to stay the execution of said judgment (docket A-42-18), a motion that FCA dismissed on April 23, 2018.\n[3] On February 6, 2018, the Applicant made his submissions concerning costs. He included an affidavit from Mr. Boulianne and filed Exhibit 1, which included three invoices and two statements of account from the firm of Neashish & Champoux s.e.n.c., indicating that he had been invoiced an amount totalling $82,544.35. On March 23, 2018, the Respondents submitted their representations concerning costs. They attached three items: the order from Prothonotary Morneau refusing the application for the Applicant’s interim costs, news articles, and the notice of appeal of the aforementioned decision dated December 21, 2017. Finally, on April 4, 2018, the Applicant submitted his response concerning costs.\n[4] The parties did not submit a bill of costs and hence the Court does not know the estimated amount of costs that would be granted according to Column III of Tariff B, if Rule 407 of the Federal Courts Rules, SOR/98-106 [the Rules] were applied.\nII. Position of the parties\n[5] Mr. Bacon St-Onge is requesting payment of costs on the attorney-client basis, thus covering all of the professional and legal fees incurred. In support of this request, he basically presented five (5) arguments, namely (1) his application for judicial review was upheld; (2) the application was brought in the public’s interest and it went beyond the scope of his individual interests; (3) unlike the Respondents, he is not in a position to have the First Nation reimburse the legal fees; (4) the case required a considerable amount of work because the facts and applicable law were complex and because the cases consisted of more than 2,000 pages; and (5) the Respondents unjustifiably refused to withdraw from a proceeding that was condemned in advance.\n[6] Mr. Bacon St-Onge also asked the Court (1) to reserve his right to again apply to a court of competent jurisdiction to claim any order and any additional sum required with respect to costs for the Respondents’ application for review; and (2) to exempt him from all the fees and expenses to be paid to the Respondents as part of this claim, the principal claim and any other ancillary or incidental claim in this case and in the appeal case.\n[7] To begin with, the Court confirms that it will not decide on these last two claims related either to possible future cases or to the appeal proceedings. Thus, this decision will be limited to the application for costs related to the litigation settled by the judgment delivered last December 21.\n[8] The Respondents reply that the expenses cannot be granted to the Applicant basically because (1) Prothonotary Morneau had refused the Applicant’s request for interim costs and there is thus res judicata on the question of expenses; and (2) the appeal dated December 21, 2017, suspends the awarding of costs and said costs will only be payable by the Applicant if their appeal is dismissed.\n[9] The Respondents add that, should costs be granted, (1) they must be calculated according to Column III of Tariff B of the Rules; (2) the questions raised in this case are not of concern to Band members, do not fall outside the individual interests of the Applicant, who showed interest in standing for election, thus showing that he had an individual interest in voiding the elections; (3) the Applicant unreasonably delayed bringing his complaint and the voters and candidates were greatly inconvenienced by the election’s invalidity; (4) the invoices that the Applicant submitted in support of his application for costs do not provide the dates and hours worked in the case and have no probative value, being domestic writings; and (5) the questions to be decided are not particularly complicated.\n[10] The Applicant replies that Prothonotary Morneau’s order decided on the application for interim costs, proceedings separate from the awarding of costs. The criteria that underlie the awarding of costs are different and, therefore, there is no res judicata in this case. Finally, the Applicant points out that he had no choice other than to turn to the courts because the Respondents refused to consider the Band members’ remarks concerning the illegality of the process for amending the 1994 Code. He thus acted for the good of all Band members. In response to the arguments concerning the format of the invoices submitted, he maintains that they are unsigned writings used in the course of business activities and that they are thus proof of their content.\n[11] Finally, the Applicant maintains that costs can be granted even if the decision is under appeal (Martselos v. Salt River Nation #195, 2008 FCA 221 at paragraphs 51 to 55).\nIII. Discussion\n[12] We should first deal with two of the arguments raised by the Respondents: the one related to the thing adjudicated and the one related to the effect of the appeal and the stay motion that were lodged.\n[13] Thus, the Court agrees with the Applicant’s position and concludes that Prothonotary Morneau’s decision on the interim costs is not res judicata on the awarding of costs at the end of the litigation. At least one of the three criteria established in Angle v. M.N.R., [1975] 2 SCR 248, the one requiring that the same question has been decided, is not satisfied here. The criteria related to a decision on the application for interim costs are different from those considered within the framework of the awarding of costs, and thus it cannot have res judicata.\n[14] As for the effect of the stay motion and the appeal that the Respondents presented to FCA, the Court notes that the Respondents did not submit any case law to support their argument. First, FCA dismissed the stay motion, and thus it is not necessary to focus on its implications with respect to the awarding of costs. Next, our Court has already agreed that appealing a Federal Court decision does not prevent the taxation of costs in the first instance (Halford v. Seed Hawk Inc., 2004 FC 1259 at paragraph 36). Thus, the Court has not been convinced that the appeal of the decision dated December 21, 2017, suspends the awarding of costs.\n[15] The Court will therefore decide on the awarding of costs and, in this regard, the Court is convinced that here, the costs must be granted in favour of the Applicant because his application for judicial review was upheld (Ticketnet Corp v. The Queen, [1999] FCA No. 1102, 99 DTC 5429).\n[16] The awarding of costs between parties is set out in sections 400 to 414 of Part II of Rules. To award costs, courts try to establish a fair balance between three principal objectives, namely “providing compensation, promoting settlement and deterring abusive behaviour” (Air Canada v. Thibodeau, 2007 FCA 115 at paragraph 24). Thus, according to Rule 407, unless the Court orders otherwise, the costs between parties are taxed in compliance with Column III of Tariff Table B.\n[17] As well, subsection 400(1) of the Rules states that the Court “shall have full discretionary power over the amount and allocation of costs and the determination of by whom they are to be paid.” The Court’s vast discretionary power over the awarding of costs has only two exceptions, related to representative actions and immigration cases, which are not at issue in this case.\n[18] Otherwise, the Court enjoys vast discretionary power (Salt River Nation #195 v. Martselos, 2008 FCA 221 at paragraphs 52 and 53). The factors that the Court may take into account are stated in subsection 400(3) of the Rules, the text of which is annexed. They include some of the factors raised by the Applicant, such as the importance and complexity of the issues (400(3)(c)), the amount of work (400(3)(g)) and whether the public interest in having the proceeding litigated justifies a particular award of costs (400(3)(h)).\n[19] The Court has the power to award a gross sum or to issue a more general order (Consorzio del Prosciutto di Parma v. Maple Leaf Meats Inc. (CA), 2002 FCA 417 at paragraphs 8 to 10).\n[20] The Court must therefore decide whether the costs will be assessed through taxation or by the awarding of a gross sum and must also decide whether there is cause to award a specific, higher amount either on the attorney-client basis or on the basis of the public interest.\n[21] To begin with, the Court rules out the payment of costs on the attorney-client basis because nothing in the case indicates that the Respondents demonstrated “reprehensible, scandalous or outrageous conduct” (Young v. Young, [1993] 4 SCR 3 at p. 134; Quebec (Attorney General) v. Lacombe, 2010 SCC 38 at paragraph 67).\n[22] As for a specific amount on the basis of public interest, the Supreme Court established, in the Carter decision (Carter v. Canada (Attorney General), 2015 SCC 5 at paragraph 140), a two-component criterion for awarding special costs to a successful party representing the public interest:\n. . . First, the case must involve matters of public interest that are truly exceptional. It is not enough that the issues raised have not previously been resolved or that they transcend the individual interests of the successful litigant: they must also have a significant and widespread societal impact. Second, in addition to showing that they have no personal, proprietary or pecuniary interest in the litigation that would justify the proceedings on economic grounds, the plaintiffs must show that it would not have been possible to effectively pursue the litigation in question with private funding. In those rare cases, it will be contrary to the interests of justice to ask the individual litigants (or, more likely, pro bono counsel) to bear the majority of the financial burden associated with pursuing the claim.\n[23] In this case, the Court notes that determining the electoral code’s validity is as much an interest for the Band as it is for the Applicant because the latter was a candidate in the elections whose cancellation he requested. Thus the Applicant cannot maintain that he had no individual interest in the litigation and here, at least one of the Supreme Court’s aforementioned criteria has not been satisfied.\n[24] In addition, it seems fair to argue that the Applicant is not in a position to get the Band to reimburse him for his legal fees. The Respondents have not submitted evidence showing that they paid their legal fees (Bellegarde v. Poitras, 2009 FC 1212 at paragraph 8) and it seems plausible to find that they are not paying them themselves, since they are members of the Band council.\n[25] Finally, the Court can find only that the workload and the complexity of the case or that the behaviour of the Respondents, having continued the proceedings, in themselves justify the awarding of special costs.\n[26] Hence, because the electoral code’s validity is effectively also a question of interest for the Band and because the Applicant is solely responsible for the litigation costs, the Court is convinced that the situation is an argument for awarding costs higher than those in Column III of Tariff B. In the absence of the parties’ bill of costs, the Court finds it difficult to set a “higher” amount by gross sum. Therefore, the Court will instead grant the Applicant costs through taxation, according to the upper band of Column V of Tariff B.\nJUDGMENT in file T-2135-16\nTHIS COURT’S JUDGMENT is that:\nThe Respondents are to pay costs to the Applicant according to the upper band of Column V of Tariff B;\n“Martine St-Louis”\nJudge\nRule 400(3)\nFactors in awarding costs\n(3) In exercising its discretion under subsection (1), the Court may consider\n(a) the result of the proceeding;\n(b) the amounts claimed and the amounts recovered;\n(c) the importance and complexity of the issues;\n(d) the apportionment of liability;\n(e) any written offer to settle;\n(f) any offer to contribute made under rule 421;\n(g) the amount of work;\n(h) whether the public interest in having the proceeding litigated justifies a particular award of costs;\n(i) any conduct of a party that tended to shorten or unnecessarily lengthen the duration of the proceeding;\n(j) the failure by a party to admit anything that should have been admitted or to serve a request to admit;\n(k) whether any step in the proceeding was\n(i) improper, vexatious or unnecessary, or\n(ii) taken through negligence, mistake or excessive caution;\n(l) whether more than one set of costs should be allowed, where two or more parties were represented by different solicitors or were represented by the same solicitor but separated their defence unnecessarily;\n(m) whether two or more parties, represented by the same solicitor, initiated separate proceedings unnecessarily;\n(n) whether a party who was successful in an action exaggerated a claim, including a counterclaim or third party claim, to avoid the operation of rules 292 to 299;\n(n.1) whether the expense required to have an expert witness give evidence was justified given\n(i) the nature of the litigation, its public significance and any need to clarify the law,\n(ii) the number, complexity or technical nature of the issues in dispute, or\n(iii) the amount in dispute in the proceeding; and\n(o) any other matter that it considers relevant.\nRègle 400(3)\nFacteurs à prendre en compte\n(3) Dans l’exercice de son pouvoir discrétionnaire en application du paragraphe (1), la Cour peut tenir compte de l’un ou l’autre des facteurs suivants :\na) le résultat de l’instance;\nb) les sommes réclamées et les sommes recouvrées;\nc) l’importance et la complexité des questions en litige;\nd) le partage de la responsabilité;\ne) toute offre écrite de règlement;\nf) toute offre de contribution faite en vertu de la règle 421;\ng) la charge de travail;\nh) le fait que l’intérêt public dans la résolution judiciaire de l’instance justifie une adjudication particulière des dépens;\ni) la conduite d’une partie qui a eu pour effet d’abréger ou de prolonger inutilement la durée de l’instance;\nj) le défaut de la part d’une partie de signifier une demande visée à la règle 255 ou de reconnaître ce qui aurait dû être admis;\nk) la question de savoir si une mesure prise au cours de l’instance, selon le cas :\n(i) était inappropriée, vexatoire ou inutile,\n(ii) a été entreprise de manière négligente, par erreur ou avec trop de circonspection;\nl) la question de savoir si plus d’un mémoire de dépens devrait être accordé lorsque deux ou plusieurs parties sont représentées par différents avocats ou lorsque, étant représentées par le même avocat, elles ont scindé inutilement leur défense;\nm) la question de savoir si deux ou plusieurs parties représentées par le même avocat ont engagé inutilement des instances distinctes;\nn) la question de savoir si la partie qui a eu gain de cause dans une action a exagéré le montant de sa réclamation, notamment celle indiquée dans la demande reconventionnelle ou la mise en cause, pour éviter l’application des règles 292 à 299;\nn.1) la question de savoir si les dépenses engagées pour la déposition d’un témoin expert étaient justifiées compte tenu de l’un ou l’autre des facteurs suivants :\n(i) la nature du litige, son importance pour le public et la nécessité de clarifier le droit,\n(ii) le nombre, la complexité ou la nature technique des questions en litige,\n(iii) la somme en litige;\no) toute autre question qu’elle juge pertinente.\nFEDERAL COURT\nSOLICITORS OF RECORD\nDOCKET:\nT-2135-16\nSTYLE OF CAUSE:\nJÉRÔME BACON ST-ONGE v. THE CONSEIL DES INNUS DE PESSAMIT, RENÉ SIMON, ÉRIC CANAPÉ, GÉRALD HERVIEUX, DIANE RIVERIN, JEAN-NOEL RIVERIN, RAYMON ROUSSELOT, MARIELLE VACHON\nREASONS FOR ORDER AND ORDER:\nST-LOUIS J.\nDATED:\nJune 22, 2018\nWRITTEN SUBMISSIONS BY:\nFrançois Boulianne\nFOR THE APPLICANT\nKenneth Gauthier\nFor the respondents\nSOLICITORS OF RECORD:\nNeashish & Champoux, s.e.n.c.\nWendake, Quebec\nFOR THE APPLICANT\nKenneth Gauthier\nCounsel\nBaie-Comeau, Quebec\nFor the respondents\n',
    "cluster: CONCLUSION:  The Court concluded that the Respondents were to pay costs to the Applicant according to the upper band of Column V of Tariff B. The Court's decision was based on its discretionary power to award costs, taking into account the factors set out in Rule 400(3) of the Federal Courts Rules. The Court's decision was intended to provide a fair balance between the parties and to reflect the complexity and importance of the issues in the case.",
    "cluster: ANALYSIS:  The Court analyzed the factors to be considered in awarding costs, as set out in Rule 400(3) of the Federal Courts Rules. The Court considered the importance and complexity of the issues, the amount of work, and the public interest in having the proceeding litigated. The Court also considered the conduct of the parties, including the Respondents' refusal to consider the Band members' remarks concerning the illegality of the process for amending the 1994 Code. The Court found that the Applicant was solely responsible for the litigation costs and that the situation justified awarding costs higher than those in Column III of Tariff B. However, the Court did not find that the case met the two-component criterion for awarding special costs to a successful party representing the public interest, as established by the Supreme Court in the Carter decision.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

<!-- ### Direct Usage (Transformers) <details><summary>Click to see the direct usage in Transformers</summary> </details> --> <!-- ### Downstream Usage (Sentence Transformers) You can finetune this model on your own dataset. <details><summary>Click to expand</summary> </details> --> <!-- ### Out-of-Scope Use *List how the model may foreseeably be misused and address what users ought not to do with the model.* --> <!-- ## Bias, Risks and Limitations *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.* --> <!-- ### Recommendations *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.* -->

Training Details

Training Dataset

Unnamed Dataset

• Size: 13,500 training samples
• Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	<ul><li>min: 410 tokens</li><li>mean: 2901.01 tokens</li><li>max: 6550 tokens</li></ul>	<ul><li>min: 18 tokens</li><li>mean: 209.23 tokens</li><li>max: 1169 tokens</li></ul>	<ul><li>min: 27 tokens</li><li>mean: 219.56 tokens</li><li>max: 1261 tokens</li></ul>

• Samples:

  anchor	positive	negative

The model card is truncated. Read the rest on Hugging Face.