metadata

language:
  - code
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - dense
  - generated_from_trainer
  - dataset_size:81143
  - loss:MultipleNegativesRankingLoss
base_model: NeuML/pubmedbert-base-embeddings
widget:
  - source_sentence: >-
      EEF1A1 FTL CD74 MALAT1 TPT1 ACTB TMSB10 WARS1 HSPA8 LCP1 EIF1 PTMA
      HSP90AB1 RBM3 FAU TYMP VIM GSTP1 CALR RACK1 TMSB4X HSP90AA1 HSPA5 SYNGR2
      STAT1 FTH1 IRF1 PPDPF BTF3 LAPTM5 HSP90B1 GDI2 WDR1 CORO1A ATP5F1E TMBIM6
      HINT1 NACA HERPUD1 MYL6 GADD45B PGK1 DDX5 GAPDH MOB1A ACTR3 CSDE1 EIF4B
      PABPC1 RBX1 ATP5F1B ARPC3 PRDX1 NCOA3 PRDX5 RAN ACTR2 SNRPG SNRNP200 ALDOA
      ATP5F1A YWHAZ PPP1CA TALDO1
    sentences:
      - >-
        This measurement was conducted with 10x 5' v1. Naive B cell from blood
        of a 26-year old male, activated with CD3.
      - >-
        EEF1A1 MALAT1 TMSB4X NACA TPT1 PABPC1 FAU PTMA FTL FTH1 NPM1 HSPA5 LDHB
        COX4I1 LCP1 SH3BGRL3 EEF2 EIF1 RACK1 GSTP1 SMCHD1 ELOB DDX5 GAPDH GTF3A
        BTF3 HNRNPU TAGLN2 RNF149 SSR2 YWHAB HNRNPF AKAP13 CALR OST4 MYCBP2 IL32
        VIM TMSB10 GABARAPL2 THRAP3 ARID4B EIF4B TRAM1 HSP90AA1 ERP29 FXYD5 EZR
        RAB18 EIF3L MYH9 EIF3E PDCD5 RABAC1 FKBP8 CHCHD2 DOCK8 HDLBP SRSF7 TMED5
        MYL12B TRIR NCOA3 EIF2S2
      - >-
        This measurement was conducted with 10x 5' v1. A 26-year-old male
        individual's blood sample, containing naive thymus-derived CD4-positive,
        alpha-beta T cells, with no activation or treatment, and in G1 phase.
  - source_sentence: >-
      MALAT1 EEF1A1 TPT1 PTMA ACTB TMSB4X H3-3B FTL FTH1 TMSB10 LGALS1 VIM CYBA
      FAU EIF1 NACA RACK1 UBA52 HSP90AA1 CD63 SH3BGRL3 LMO4 HMGB1 S100A4 UBC
      HNRNPU HSP90AB1 DDX5 DUSP1 HNRNPA2B1 SOX4 JUND DBI S100A6 GSTP1 MYL6 PFN1
      GAPDH SRGN SERF2 TAGLN2 IER2 UBB CFL1 JUN YBX1 PABPC1 OAZ1 ARPC3 CCNI DAD1
      BTG1 ATP5MC2 BTF3 ZFP36L2 TSC22D3 EEF2 FOS IFITM2 PPIA KLF6 GNAS DYNLL1
      MYL12A
    sentences:
      - >-
        This measurement was conducted with 10x 3' v3. Blasts cells derived from
        the blood of a 4-month old male.
      - >-
        MALAT1 TPT1 HSP90B1 SSR4 SUB1 EEF1A1 SAT1 XBP1 SPCS1 ITM2C PPIB SEC61B
        TMBIM6 SEC61G CYBA FAU UBC NACA SELENOS TMSB10 SEC11C UBE2J1 CALR TXNIP
        HSPA5 ACTB SELENOK SPCS2 RRBP1 UBA52 H3-3B SERF2 FTH1 EIF1 SEC62 NUCB2
        SSR2 VIM ERLEC1 MYL6 SRGN ATP5F1E PTMA NEAT1 TRAM1 GNAS KLF6 LMAN1 MYDGF
        TMEM59 IFI6 ARPC2 H1-10 CD74 HERPUD1 HSP90AA1 OAZ1 GAPDH SSR3 CCNI SPCS3
        COX4I1 ITM2B TXN
      - >-
        This measurement was conducted with 10x 3' v3. This is a
        megakaryocyte-erythroid progenitor cell (MEP-like) derived from a
        1-month-old female patient with KMT2A-rearranged (KMT2A-r) infant acute
        lymphoblastic leukemia (ALL). The cell exhibits increased lineage
        plasticity, downregulated steroid response pathways, and belongs to a
        hematopoietic stem and progenitor-like (HSPC-like) population that forms
        an immunosuppressive signaling circuit with cytotoxic lymphocytes.
  - source_sentence: >-
      MALAT1 NRXN3 NRXN1 DPP10 ADARB2 IL1RAPL1 CADM2 MEIS2 ROBO2 PBX1 SOX2-OT
      CELF2 RALYL DPP6 PBX3 KALRN SLC8A1 MEG3 EPHA6 GRIK1 ERBB4 SPOCK3 ENOX1
      ZNF385D KCND2 ADCY2 PDE4D GRIP1 TCF4 SNTG1 NTRK2 PCDH9 NKAIN2 MAML3 DAB1
      GRIA1 LRFN5 NTM EPHA5 ANK3 LSAMP MAP2 DCLK1 TRPM3 KIRREL3 PPM1E KCNQ5
      SIPA1L1 CHSY3 RORA CACNA2D1 CADPS TAFA2 KLHL29 TIAM1 GABRB3 ROBO1 FUT9
      ATRNL1 FGF13 NCAM1 AKAP6 L3MBTL4 IL1RAPL2
    sentences:
      - >-
        MALAT1 PLP1 PCDH9 IL1RAPL1 PTPRD QKI ST18 MAN2A1 KIRREL3 PDE4B GPM6B
        ANK3 EDIL3 MOB3B MBP PHLPP1 MAP7 TMEFF2 PPP2R2B ZEB2 MAML3 SLC44A1 PTGDS
        FOXP1 DOCK4 SLC24A2 MAP4K5 SGK1 APP DOCK5 ELMO1 FMNL2 SIK3 FRMD5 SHTN1
        GRM3 LINC00609 PICALM APLP1 DNAJC6 MACF1 TMEM165 EXOC6B HHIP YPEL2
        CTNNA3 SOX2-OT DBNDD2 CD22 VRK2 FUT8 PLEKHH1 ANKRD44 SLCO1A2 COBL
        ARHGAP21 CCDC88A FOXO3 ATP8A1 HIP1 ENPP2 PPM1B SECISBP2L IGF1R
      - >-
        This measurement was conducted with 10x 3' v3. Neuron cell type from a
        50-year-old male human cerebral cortex, specifically from the Cingulate
        gyrus, rostral (CgGr), Ventral division of MFC - A24 region, with
        European self-reported ethnicity, analyzed at the nucleus level.
      - >-
        This measurement was conducted with 10x 3' v3. Neuron cell type from a
        50-year-old male human cerebral cortex, specifically the rostral
        cingulate gyrus, ventral division of MFC, A24, with European ethnicity.
  - source_sentence: >-
      MALAT1 TPT1 SSR4 HSP90AA1 EEF1A1 JUN KLF6 FTL FOS BTG2 SAT1 JUNB PPIB CD74
      XBP1 DUSP1 SEC11C RGCC UBC SERF2 HSP90B1 HERPUD1 FAU TSC22D3 CYBA HM13
      SERP1 NEAT1 CD38 TMBIM6 RPN1 PSAP OST4 TMSB10 LMAN1 SEC61B RRBP1 DNAJB1
      RHOB EIF1 UBE2J1 HSPA5 SSR3 KLF2 P4HB MYDGF SPCS2 ITM2C UBB TMED9 SEL1L
      SUB1 SPCS1 SEC61G MCL1 FTH1 CALR RABAC1 COX7A2 NCL RAB30 PABPC1 SEL1L3
      KDELR1
    sentences:
      - >-
        This measurement was conducted with 10x 5' v2. Memory B cell derived
        from a 65-79 year-old male, taken from the mesenteric lymph node.
      - >-
        This measurement was conducted with 10x 5' v2. IgA plasma cell sample
        taken from the mesenteric lymph node of a 65-79 year-old female.
      - >-
        EEF1A1 TPT1 TMSB4X ACTB MALAT1 FOS DUSP1 KLF2 FAU JUNB PTMA TMSB10
        DNAJB1 FTL JUN NACA FTH1 TSC22D3 EIF1 PFN1 HSPA8 LDHB H3-3B BTG1 ZFP36L2
        NPM1 IL32 VIM PABPC1 CORO1A COX4I1 BTF3 UBC DUSP2 EEF2 EEF1G ARHGEF1
        HSP90AB1 CIRBP MYL12A NR4A1 ZFP36 ANXA1 ITM2B NOSIP PNRC1 UQCRB BTG2
        LAPTM5 PCBP1 COMMD6 S100A4 PPIA UBA52 CD44 FAM107B YBX1 HSP90AA1 GAPDH
        HSPE1 SRSF7 SERP1 CXCR4 PPDPF
  - source_sentence: >-
      EEF1A1 TMSB4X MALAT1 CD74 H3-3B FAU TPT1 ACTB FTH1 PTMA EIF1 ZFP36L1 UBA52
      NPM1 PPIA HSP90AA1 RGS2 SAT1 TSC22D3 EEF2 HMGB1 GRN STK17B COTL1 EDF1 CD83
      PRDX1 ZFP36 COX4I1 ANP32B EML4 TAF1D UQCRH NACA RACK1 ENO1 RBM3 PFN1 PARK7
      IRF1 SNRPD2 SNRPB COX7C KLF2 ATP6V1F ZNF331 BTF3 EIF3H HNRNPDL UQCRB
      EIF4A2 TAGLN2 ARPC2 YWHAB SF1 EIF3F ZFAS1 H4C3 TMSB10 HERPUD1 SLC2A3 WNK1
      MEF2A ARHGAP15
    sentences:
      - >-
        MALAT1 CD74 EEF1A1 ACTB EIF1 TMSB4X PTMA TSC22D3 FTL TPT1 BTG1 FTH1 UBC
        TMSB10 KLF6 FAU PNRC1 HSP90AB1 CD83 LAPTM5 JUN NACA RACK1 HLA-DRB5 DDX5
        KLF2 IRF8 GPR183 TXNIP PPP1R15A NFKBIA YPEL5 H3-3B ZFP36L2 YWHAZ UBA52
        CYBA OAZ1 DUSP1 SARAF RHOA MYL12A COTL1 PFN1 HSPA8 MCL1 TAGLN2 TUBA1A
        CALM1 HMGN2 BCLAF1 PABPC1 HSP90AA1 SMAP2 EZR ARPC3 ACTR3 EPC1 CXCR4
        SEPTIN7 ZFP36 SNX9 EEF2 FOS
      - >-
        This measurement was conducted with 10x 5' v1. Memory B cell derived
        from a 3-year-old male human tonsil tissue, expressing IGHJ4*02,
        IGHV4-59*01, IGKV3-20, IGKJ2, and IgG1 isotype.
      - >-
        This measurement was conducted with 10x 5' v1. Plasmablast cell sample
        from a 3-year-old male, taken from the tonsil tissue, expressing IgM
        isotype, with IGH_IN_FRAME, IGH_FUNCTIONAL, IGH_JUNCTION_LENGTH 48.0,
        IGH_J_CALL IGHJ3*02, IGH_V_CALL_GENOTYPED IGHV4-39*01, IGK_C_Gene IGKC,
        IGK_FullLength 2, IGK_Productive 2, IGK_VDJ_Gene IGKV3-20 None IGKJ1.
datasets:
  - jo-mengr/cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy
model-index:
  - name: SentenceTransformer based on NeuML/pubmedbert-base-embeddings
    results:
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: >-
            cellxgene pseudo bulk 100k multiplets natural language annotation
            cell sentence 2
          type: >-
            cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation_cell_sentence_2
        metrics:
          - type: cosine_accuracy
            value: 0.4768616259098053
            name: Cosine Accuracy

SentenceTransformer based on NeuML/pubmedbert-base-embeddings

This is a sentence-transformers model finetuned from NeuML/pubmedbert-base-embeddings on the cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation dataset. 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: NeuML/pubmedbert-base-embeddings
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 dimensions
Similarity Function: Cosine Similarity
Training Dataset:
- cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation
Language: code

Model Sources

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

Full Model Architecture

SentenceTransformer(
  (0): MMContextEncoder(
    (text_encoder): BertModel(
      (embeddings): BertEmbeddings(
        (word_embeddings): Embedding(30522, 768, padding_idx=0)
        (position_embeddings): Embedding(512, 768)
        (token_type_embeddings): Embedding(2, 768)
        (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
        (dropout): Dropout(p=0.1, inplace=False)
      )
      (encoder): BertEncoder(
        (layer): ModuleList(
          (0-11): 12 x BertLayer(
            (attention): BertAttention(
              (self): BertSdpaSelfAttention(
                (query): Linear(in_features=768, out_features=768, bias=True)
                (key): Linear(in_features=768, out_features=768, bias=True)
                (value): Linear(in_features=768, out_features=768, bias=True)
                (dropout): Dropout(p=0.1, inplace=False)
              )
              (output): BertSelfOutput(
                (dense): Linear(in_features=768, out_features=768, bias=True)
                (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
                (dropout): Dropout(p=0.1, inplace=False)
              )
            )
            (intermediate): BertIntermediate(
              (dense): Linear(in_features=768, out_features=3072, bias=True)
              (intermediate_act_fn): GELUActivation()
            )
            (output): BertOutput(
              (dense): Linear(in_features=3072, out_features=768, bias=True)
              (LayerNorm): LayerNorm((768,), eps=1e-12, elementwise_affine=True)
              (dropout): Dropout(p=0.1, inplace=False)
            )
          )
        )
      )
      (pooler): BertPooler(
        (dense): Linear(in_features=768, out_features=768, bias=True)
        (activation): Tanh()
      )
    )
    (pooling): 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("jo-mengr/mmcontext-pubmedbert-100k-v3")
# Run inference
sentences = [
    'EEF1A1 TMSB4X MALAT1 CD74 H3-3B FAU TPT1 ACTB FTH1 PTMA EIF1 ZFP36L1 UBA52 NPM1 PPIA HSP90AA1 RGS2 SAT1 TSC22D3 EEF2 HMGB1 GRN STK17B COTL1 EDF1 CD83 PRDX1 ZFP36 COX4I1 ANP32B EML4 TAF1D UQCRH NACA RACK1 ENO1 RBM3 PFN1 PARK7 IRF1 SNRPD2 SNRPB COX7C KLF2 ATP6V1F ZNF331 BTF3 EIF3H HNRNPDL UQCRB EIF4A2 TAGLN2 ARPC2 YWHAB SF1 EIF3F ZFAS1 H4C3 TMSB10 HERPUD1 SLC2A3 WNK1 MEF2A ARHGAP15',
    "This measurement was conducted with 10x 5' v1. Plasmablast cell sample from a 3-year-old male, taken from the tonsil tissue, expressing IgM isotype, with IGH_IN_FRAME, IGH_FUNCTIONAL, IGH_JUNCTION_LENGTH 48.0, IGH_J_CALL IGHJ3*02, IGH_V_CALL_GENOTYPED IGHV4-39*01, IGK_C_Gene IGKC, IGK_FullLength 2, IGK_Productive 2, IGK_VDJ_Gene IGKV3-20 None IGKJ1.",
    "This measurement was conducted with 10x 5' v1. Memory B cell derived from a 3-year-old male human tonsil tissue, expressing IGHJ4*02, IGHV4-59*01, IGKV3-20, IGKJ2, and IgG1 isotype.",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 1.0000, 1.0000],
#         [1.0000, 1.0000, 1.0000],
#         [1.0000, 1.0000, 1.0000]])

Evaluation

Metrics

Triplet

Dataset: cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation_cell_sentence_2
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.4769

Training Details

Training Dataset

cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation

Dataset: cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation at 7041d95
Size: 81,143 training samples
Columns: anchor, positive, negative_1, and negative_2

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative_1	negative_2
type	string	string	string	string
details	min: 356 characters mean: 385.24 characters max: 450 characters	min: 92 characters mean: 216.13 characters max: 900 characters	min: 101 characters mean: 215.14 characters max: 870 characters	min: 338 characters mean: 384.66 characters max: 433 characters

Samples:

anchor	positive	negative_1	negative_2
`TMSB4X TMSB10 ACTB MALAT1 GNLY NKG7 IFITM2 LGALS1 GZMA EEF1A1 PFN1 HMGB2 FTH1 PTMA HSP90AA1 GZMB ARHGDIB HNRNPA2B1 PLAAT4 FAU CMC1 VIM MYL12A CBX3 ATP5F1E HCST IFI44L KLRF1 H3-3A COX6C ARL6IP1 CFL1 ISG15 HMGB1 S100A4 ATP5MF RORA MYL6 CORO1A OAZ1 KLRB1 ID2 HMGN3 CCNI RBM39 CAP1 SERF2 ELOC FCER1G S100A9 IFI16 YWHAZ EIF1 CALR HMGN2 SKAP2 SLC25A5 ZZZ3 YBX1 NUCB2 CDC42 GSTP1 FTL ATP5F1D`	`This measurement was conducted with 10x 3' v2. A proliferating lymphocyte cell sample, obtained from a 34-year-old female Asian individual, derived from peripheral blood mononuclear cells.`	`This measurement was conducted with 10x 3' v2. Sample is a 25-year-old female with European ethnicity, having CD8-positive, alpha-beta T cell type. This cell type exhibits elevated expression of type 1 interferon-stimulated genes (ISGs) in monocytes, reduction of naïve CD4+ T cells correlating with monocyte ISG expression, and expansion of repertoire-restricted cytotoxic GZMH+ CD8+ T cells.`	`MALAT1 TMSB4X EEF1A1 CD74 BTG1 PTMA TMSB10 TPT1 FAU EIF1 FTH1 FTL CXCR4 TSC22D3 DUSP1 UBA52 ACTB CD37 CD52 NACA RACK1 EZR CD69 LAPTM5 H3-3A FOS ISG20 YBX1 CIRBP EIF3E OAZ1 COX7C SAT1 COX4I1 H3-3B SH3BGRL3 UBC UBB JUNB COMMD6 VIM CYBA KLF6 STK17B FUS HNRNPC MYL6 GADD45B LGALS1 EIF3L SRSF5 NFKBIA ANKRD12 CORO1A TLE5 NOP53 CHCHD2 PFN1 DDX5 ARPC3 COX7A2 YPEL5 ARL4A SRGN`
`EEF1A1 MALAT1 FTH1 JUNB TPT1 FOS TMSB10 BTG1 TMSB4X ZFP36L2 NACA PABPC1 ACTB FAU VIM H3-3B EIF1 ZFP36 SARAF PTMA IL7R JUN RACK1 EEF2 UBA52 GAPDH FTL FXYD5 DUSP1 S100A4 CD69 CXCR4 UBC TSC22D3 CFL1 KLF6 ARHGDIB KLF2 BTG2 CITED2 IER2 TUBB4B CD3E EEF1G SLC2A3 NFKBIA PFN1 SRGN SNX9 COX4I1 DNAJB1 SERF2 CD8A PCBP2 IL32 BIRC3 SMAP2 FUS GADD45B MYL12A OAZ1 ATP5F1E TUBA4A PNRC1`	`This measurement was conducted with 10x 5' v1. Sample is a cell from the omentum tissue, specifically an effector memory CD4-positive, alpha-beta T cell, from a female in her sixth decade.`	`This measurement was conducted with 10x 5' v2. Conventional dendritic cell from the jejunal epithelium of a female in her eighth decade.`	`CD74 MALAT1 EEF1A1 FOS TPT1 TMSB4X TMSB10 ACTB FAU JUN CD37 DUSP1 RACK1 JUNB EIF1 PTMA FTL DNAJB1 H3-3B CD52 NACA BTG1 TSC22D3 FTH1 PABPC1 EEF2 UBA52 EEF1G HSP90AA1 LAPTM5 CYBA PPP1R15A HSP90AB1 CD69 ARHGDIB ZFP36 SERF2 UBC H3-3A PCBP2 HLA-DRB5 KLF6 PFN1 DDX5 HSPA8 ARPC3 CD83 CCNI CXCR4 ATP5F1E SARAF TUBA1A ZFP36L1 TOMM7 HERPUD1 YBX1 RHOA MEF2C FXYD5 MYL6 SRSF5 MYL12A CORO1A OAZ1`
`MALAT1 GRIK1 SYT1 PCDH9 RORA NRG1 CADPS ZFPM2 LRRC4C LINGO2 RALYL PTPRD SPHKAP CNTNAP5 SLC8A1 CCSER1 HDAC9 CELF2 R3HDM1 CNTN4 RBMS3 PCDH7 GALNT13 UNC5D ROBO1 SYNPR SNAP25 GPM6A ANK3 FRMPD4 CHRM2 RYR2 KHDRBS2 CADM1 CACNA1D RGS6 PDE4D DOCK4 UNC13C CDH18 FAT3 MEG3 NR2F2-AS1 HMCN1 GULP1 CAMK2D ZEB1 SYN2 DYNC1I1 OXR1 DPP10 OSBPL6 FRAS1 PPP3CA ZNF385D ZMAT4 PCBP3 HS6ST3 ERC2 PLEKHA5 CDK14 MAP2 NCOA1 ATP8A2`	`This measurement was conducted with 10x 3' v3. Neuron cell type from a 29-year-old male, specifically from the thalamic complex, specifically the thalamus (THM) - posterior nuclear complex of thalamus (PoN) - medial geniculate nuclei (MG).`	`This measurement was conducted with 10x 3' v3. Neuron from the thalamic complex (thalamus, posterior nuclear complex of thalamus, medial geniculate nuclei) of a 42-year-old male, identified as a midbrain-derived inhibitory neuron.`	`MALAT1 PCDH9 PTPRD NRG1 SYT1 DPP10 ROBO1 TENM2 LRRC4C RBMS3 CNTNAP5 LINGO2 CDH18 SLC8A1 DMD PDE4D RYR2 ATP1B1 RGS6 PTPRT CHRM3 ADGRL2 NOVA1 NTNG1 PCDH7 TAFA2 CCSER1 ANK3 MEG3 MAP2 PLCB4 CACNA2D1 PRKG1 LINC03000 RMST RORA FOXP2 LHFPL3 MEG8 TNRC6A DAB1 KCTD8 RALYL GNAS INPP4B OLFM3 CNTN4 FRMD4A LINC00632 GAPDH ENOX1 AHI1 GPM6A EBF1 LRFN5 PCSK1N SEMA5A KIAA1217 CALY MAP1B SNAP25 GABRB2 CDH8 GRIP1`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Evaluation Dataset

cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation

Dataset: cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation at 7041d95
Size: 9,011 evaluation samples
Columns: anchor, positive, negative_1, and negative_2

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative_1	negative_2
type	string	string	string	string
details	min: 347 characters mean: 386.7 characters max: 437 characters	min: 99 characters mean: 209.99 characters max: 941 characters	min: 102 characters mean: 213.87 characters max: 981 characters	min: 347 characters mean: 386.42 characters max: 433 characters

Samples:

anchor	positive	negative_1	negative_2
`MALAT1 EEF1A1 FTH1 TMSB4X ACTB FTL RTN4 ATP6V0B TPT1 FAU S100A6 NDUFA4 ATP5F1E COX7C ITM2B IGFBP7 EIF1 C12orf75 CD9 COX7B SERF2 ATP1B1 COX8A TXNIP NDUFB2 MYL6 PPDPF COX6B1 UQCR11 APOE COX4I1 CALM2 UQCRB S100A11 UQCRQ COX6C ATP5MG BSG ATP6AP2 UQCR10 PTMA NACA UBL5 UBA52 TMSB10 ADGRF5 HSP90AA1 GSTP1 ATP5F1D CHCHD2 GAPDH COX7A2 SKP1 HSPE1 PRDX1 CYSTM1 LGALS3 CD63 ATP5MJ CKB NDUFS5 ATP5ME UBB MAL`	`This measurement was conducted with 10x 3' v3. Cell sample from the cortex of kidney, taken from a 43-year-old male of European ethnicity with a reported history of kidney cancer. The cell type is identified as a kidney collecting duct intercalated cell.`	`This measurement was conducted with 10x 3' v3. Kidney collecting duct intercalated cell from a 43-year old European male with kidney cancer, taken from the cortex of kidney and cryopreserved for further analysis.`	`MALAT1 EEF1A1 CRYAB S100A6 ITM2B ACTB TPT1 PTMA FTL PEBP1 H3-3B GSTP1 ADIRF IGFBP7 S100A10 HIPK2 MYL6 SERF2 TPM1 FAU FTH1 ID4 EIF1 TMSB10 HSP90AA1 SKP1 IGFBP2 IGFBP5 PRDX1 MYL12B CYSTM1 CLU ATP5F1E AHNAK PPDPF DSTN ID1 COX7C JUND SRP14 ATP1B1 HINT1 NDUFA4 PPIA NACA TMA7 NEAT1 CD9 SYNE2 LAPTM4A GNAS CIRBP ATP5F1D DDX17 EDF1 CCND1 LDHB RTN4 TMEM59 NR4A1 KTN1 SAT1 TMBIM6 APP`
`MALAT1 KCND2 NRXN1 CDH18 NRXN3 ZNF385D CADM2 RALYL NKAIN2 CADPS2 RIMS1 FSTL5 GRID2 TRPM3 CHN2 DPP6 JMJD1C RORA PDE1A UNC13C TIAM1 NRG1 SNAP25 ZFPM2 CALN1 LSAMP CNTN1 ABLIM1 SYNE1 ANK3 CA10 NFIA ZBTB20 NTM CADM1 OPCML RELN DNM3 NEBL ERC1 SCN2A PPP3CA CACNA1A GALNT13 LRRC4C GPM6A RABGAP1L RIT2 CAMK4 GRIA4 PTPRD RBFOX3 MCTP1 LHFPL6 PCLO MEG3 PDE10A NOVA1 RTN1 ZNF385B CNTN4 GABRB2 SPOCK1 OXR1`	`This measurement was conducted with 10x 3' v3. Neuron cell type from a 29-year-old male cerebellum, specifically from the Cerebellar Vermis - CBV region, with European self-reported ethnicity, analyzed at the nucleus level.`	`This measurement was conducted with 10x 3' v3. Endothelial cells derived from the cerebellum (specifically, cerebellar vermis) of a 42-year-old male, classified under the vascular supercluster term.`	`MALAT1 ATP10A COBLL1 GPCPD1 PTPRG SLC39A10 FLT1 FLI1 TSPAN5 THSD4 RUNDC3B CCNY IGFBP7 ST6GALNAC3 PRKCH ST6GAL1 MECOM ESYT2 TBC1D4 IGF1R TACC1 HERC4 CDH2 TCF4 ABCB1 DOCK9 SORBS2 USP54 CBFA2T2 TSC22D1 QKI EPAS1 APP NFIB AOPEP ELMO1 ZNF704 PTPRM NET1 A2M FGD6 EPHA3 NEBL RAPGEF2 ACVR1 SPTBN1 BBS9 KLF2 MKLN1 EXOC6 LEF1 PPP3CA RBMS3 LRMDA WDFY3 BCL2L1 TTC3 SIPA1L1 CFLAR ADGRF5 MAP4K4 SCARB1 RAPGEF4 ABLIM1`
`EEF1A1 ACTB GAPDH HMGN2 PTMA SERF2 TMSB4X CD74 PABPC1 FTH1 TMSB10 FAU PFN1 HMGN1 OAZ1 HMGB1 TPT1 PPIA NACA BTF3 MALAT1 MYL6 ATP5MG CFL1 RACK1 ODC1 ATP5F1E TMA7 SLC25A5 ELOB ARPC3 NPM1 COX7C ANP32B C4orf3 EIF1 PCBP2 KLF6 LAPTM5 COX8A RHOA HSPA8 H3-3B PTP4A2 UBA52 OST4 CIRBP LGALS1 EIF3L STMN1 PPDPF COX4I1 RAN EIF3F PPP1CC COMMD6 NDUFA4 YBX1 PEBP1 COTL1 COX7A2 HSPE1 CCNI TRIR`	`This measurement was conducted with 10x 5' v1. Cell sample from the tonsil of a 9-year-old female with recurrent tonsillitis, characterized as a centroblast B cell with IGLC2, IGLV7-43, IGLJ3 immunoglobulin genes expressed.`	`This measurement was conducted with 10x 5' v1. Centroblast cells derived from a 3-year-old male human tonsil sample, with obstructive sleep apnea and recurrent tonsillitis, undergoing affinity maturation and differentiation into memory or plasma cells.`	`CD74 MALAT1 EEF1A1 ACTB TMSB4X LAPTM5 PTMA TPT1 TMSB10 CXCR4 FAU BTG1 TXNIP PABPC1 FTH1 NACA FTL IRF1 RBM3 CD83 CCNI SARAF BTF3 HNRNPA3 HLA-DRB5 UBA52 MEF2C CORO1A UBE2D3 ATP5F1E PDIA6 UBC GABARAP CFL1 CALR RACK1 HSPA5 EIF4B RHOA HNRNPC SRSF5 PFN1 HSPA8 CNOT2 IFT57 HNRNPA2B1 COX7C ITM2B SH3BGRL3 PNRC1 PDIA3 EEF2 UBB PARP14 SNX2 LAP3 SLC25A5 POU2F2 ADAM28 ZNF800 CYBA GDI2 STK17B EIF3I`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 128
per_device_eval_batch_size: 128
learning_rate: 0.05
num_train_epochs: 4
warmup_ratio: 0.1
bf16: True

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 128
per_device_eval_batch_size: 128
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 0.05
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 4
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.1
warmup_steps: 0
log_level: passive
log_level_replica: warning
log_on_each_node: True
logging_nan_inf_filter: True
save_safetensors: True
save_on_each_node: False
save_only_model: False
restore_callback_states_from_checkpoint: False
no_cuda: False
use_cpu: False
use_mps_device: False
seed: 42
data_seed: None
jit_mode_eval: False
use_ipex: False
bf16: True
fp16: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: 0
ddp_backend: None
tpu_num_cores: None
tpu_metrics_debug: False
debug: []
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_prefetch_factor: None
past_index: -1
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: False
ignore_data_skip: False
fsdp: []
fsdp_min_num_params: 0
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
fsdp_transformer_layer_cls_to_wrap: None
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch
optim_args: None
adafactor: False
group_by_length: False
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
dataloader_pin_memory: True
dataloader_persistent_workers: False
skip_memory_metrics: True
use_legacy_prediction_loop: False
push_to_hub: False
resume_from_checkpoint: None
hub_model_id: None
hub_strategy: every_save
hub_private_repo: None
hub_always_push: False
hub_revision: None
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
include_for_metrics: []
eval_do_concat_batches: True
fp16_backend: auto
push_to_hub_model_id: None
push_to_hub_organization: None
mp_parameters:
auto_find_batch_size: False
full_determinism: False
torchdynamo: None
ray_scope: last
ddp_timeout: 1800
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
include_tokens_per_second: False
include_num_input_tokens_seen: False
neftune_noise_alpha: None
optim_target_modules: None
batch_eval_metrics: False
eval_on_start: False
use_liger_kernel: False
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	cellxgene pseudo bulk 100k multiplets natural language annotation loss	cellxgene_pseudo_bulk_100k_multiplets_natural_language_annotation_cell_sentence_2_cosine_accuracy
0.1577	100	5.7909	5.3928	0.5671
0.3155	200	5.7198	5.9454	0.4392
0.4732	300	5.9533	5.9454	0.4067
0.6309	400	5.9509	5.9454	0.3111
0.7886	500	5.9507	5.9454	0.3115
0.9464	600	5.9507	5.9454	0.2933
1.1041	700	5.9499	5.9454	0.3154
1.2618	800	5.9507	5.9454	0.4052
1.4196	900	5.9508	5.9454	0.4603
1.5773	1000	5.9511	5.9455	0.4445
1.7350	1100	5.9513	5.9455	0.4450
1.8927	1200	5.9512	5.9455	0.4710
2.0505	1300	5.9506	5.9455	0.4730
2.2082	1400	5.9517	5.9455	0.4721
2.3659	1500	5.9517	5.9455	0.4705
2.5237	1600	5.9517	5.9455	0.4723
2.6814	1700	5.9517	5.9455	0.4751
2.8391	1800	5.9517	5.9455	0.4722
2.9968	1900	5.9517	5.9455	0.4729
3.1546	2000	5.9509	5.9455	0.4677
3.3123	2100	5.9517	5.9455	0.4693
3.4700	2200	5.9517	5.9455	0.4728
3.6278	2300	5.9517	5.9455	0.4702
3.7855	2400	5.9517	5.9455	0.4718
3.9432	2500	5.9517	5.9455	0.4769

Framework Versions

Python: 3.11.6
Sentence Transformers: 5.0.0
Transformers: 4.55.0.dev0
PyTorch: 2.5.1+cu121
Accelerate: 1.9.0
Datasets: 2.19.1
Tokenizers: 0.21.4

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}