metadata

base_model: pritamdeka/assamese-bert-nli-v2
datasets: []
language: []
library_name: sentence-transformers
metrics:
  - pearson_cosine
  - spearman_cosine
  - pearson_manhattan
  - spearman_manhattan
  - pearson_euclidean
  - spearman_euclidean
  - pearson_dot
  - spearman_dot
  - pearson_max
  - spearman_max
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:5749
  - loss:CosineSimilarityLoss
widget:
  - source_sentence: >-
      আমি "... comoving মহাজাগতিক বিশ্ৰাম ফ্ৰেমৰ তুলনাত ... সিংহ নক্ষত্ৰমণ্ডলৰ
      ফালে কিছু 371 কিলোমিটাৰ প্ৰতি ছেকেণ্ডত" আগবাঢ়িছো.
    sentences:
      - বাস্কেটবল খেলুৱৈগৰাকীয়ে নিজৰ দলৰ হৈ পইণ্ট লাভ কৰিবলৈ ওলাইছে।
      - আন কোনো বস্তুৰ লগত আপেক্ষিক নহোৱা কোনো ‘ষ্টিল’ নাই।
      - এজনী ছোৱালীয়ে বতাহ বাদ্যযন্ত্ৰ বজায়।
  - source_sentence: চাৰিটা ল’ৰা-ছোৱালীয়ে ভঁৰালৰ জীৱ-জন্তুবোৰলৈ চাই আছে।
    sentences:
      - ডাইনিং টেবুল এখনৰ চাৰিওফালে বৃদ্ধৰ দল এটাই পোজ দিছে।
      - বিকিনি পিন্ধা চাৰিগৰাকী মহিলাই বিলত ভলীবল খেলি আছে।
      - ল’ৰা-ছোৱালীয়ে ভেড়া চাই।
  - source_sentence: ডালত বহি থকা দুটা টান ঈগল।
    sentences:
      - জাতৰ জেব্ৰা ডানিঅ’ অত্যন্ত কঠোৰ মাছ, ইহঁতক হত্যা কৰাটো প্ৰায় কঠিন।
      - এটা ডালত দুটা ঈগল বহি আছে।
      - >-
        নূন্যতম মজুৰিৰ আইনসমূহে কম দক্ষ, কম উৎপাদনশীল লোকক আটাইতকৈ বেছি আঘাত
        দিয়ে।
  - source_sentence: >-
      "মই আচলতে যি বিচাৰিছো সেয়া হৈছে মুছলমান জনসংখ্যাৰ এটা অনুমান..." @ThanosK
      আৰু @T.E.D., এটা সামগ্ৰিক, সাধাৰণ জনসংখ্যাৰ অনুমান f.e.
    sentences:
      - এগৰাকী মহিলাই সেউজীয়া পিঁয়াজ কাটি আছে।
      - >-
        তলত দিয়া কথাখিনি মোৰ কুকুৰ কাণৰ দৰে কপিৰ পৰা লোৱা হৈছে নিউ পেংগুইন
        এটলাছ অৱ মেডিভেল হিষ্ট্ৰীৰ।
      - আমাৰ দৰে সৌৰজগতৰ কোনো তাৰকাৰাজ্যৰ বাহিৰত থকাটো সম্ভৱ হ’ব পাৰে।
  - source_sentence: ইণ্টাৰনেট কেমেৰাৰ জৰিয়তে এগৰাকী ছোৱালীৰ লগত কথা পাতিলে মানুহজনে।
    sentences:
      - গছৰ শাৰী এটাৰ সন্মুখত পথাৰত ভেড়া চৰিছে।
      - এজন মানুহে গীটাৰ বজাই আছে।
      - ৱেবকেমৰ জৰিয়তে এগৰাকী ছোৱালীৰ সৈতে কথা পাতিছে এজন কিশোৰে।
model-index:
  - name: SentenceTransformer based on pritamdeka/assamese-bert-nli-v2
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: pritamdeka/stsb assamese translated dev
          type: pritamdeka/stsb-assamese-translated-dev
        metrics:
          - type: pearson_cosine
            value: 0.8582086169969396
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.8558833817052474
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.8402288134127139
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.8466669319881411
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.8401702610820984
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.846937443225358
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.8293854931734366
            name: Pearson Dot
          - type: spearman_dot
            value: 0.8279065905764471
            name: Spearman Dot
          - type: pearson_max
            value: 0.8582086169969396
            name: Pearson Max
          - type: spearman_max
            value: 0.8558833817052474
            name: Spearman Max
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: pritamdeka/stsb assamese translated test
          type: pritamdeka/stsb-assamese-translated-test
        metrics:
          - type: pearson_cosine
            value: 0.8231106499789409
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.8235370017309012
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.8131384280231726
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.817044158823682
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.8132779879142208
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.8170404249477559
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.7896666837864712
            name: Pearson Dot
          - type: spearman_dot
            value: 0.7870703093898731
            name: Spearman Dot
          - type: pearson_max
            value: 0.8231106499789409
            name: Pearson Max
          - type: spearman_max
            value: 0.8235370017309012
            name: Spearman Max

SentenceTransformer based on pritamdeka/assamese-bert-nli-v2

This is a sentence-transformers model finetuned from pritamdeka/assamese-bert-nli-v2. 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: pritamdeka/assamese-bert-nli-v2
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 tokens
Similarity Function: Cosine Similarity

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': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (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("pritamdeka/assamese-bert-nli-v2-assamese-sts")
# Run inference
sentences = [
    'ইণ্টাৰনেট কেমেৰাৰ জৰিয়তে এগৰাকী ছোৱালীৰ লগত কথা পাতিলে মানুহজনে।',
    'ৱেবকেমৰ জৰিয়তে এগৰাকী ছোৱালীৰ সৈতে কথা পাতিছে এজন কিশোৰে।',
    'এজন মানুহে গীটাৰ বজাই আছে।',
]
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]

Evaluation

Metrics

Semantic Similarity

Dataset: pritamdeka/stsb-assamese-translated-dev
Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8582
spearman_cosine	0.8559
pearson_manhattan	0.8402
spearman_manhattan	0.8467
pearson_euclidean	0.8402
spearman_euclidean	0.8469
pearson_dot	0.8294
spearman_dot	0.8279
pearson_max	0.8582
spearman_max	0.8559

Semantic Similarity

Dataset: pritamdeka/stsb-assamese-translated-test
Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8231
spearman_cosine	0.8235
pearson_manhattan	0.8131
spearman_manhattan	0.817
pearson_euclidean	0.8133
spearman_euclidean	0.817
pearson_dot	0.7897
spearman_dot	0.7871
pearson_max	0.8231
spearman_max	0.8235

Training Details

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
num_train_epochs: 4
warmup_ratio: 0.1
fp16: 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: 16
per_device_eval_batch_size: 16
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_steps: None
learning_rate: 5e-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: False
fp16: True
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: False
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
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
dispatch_batches: None
split_batches: 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
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	pritamdeka/stsb-assamese-translated-dev_spearman_cosine	pritamdeka/stsb-assamese-translated-test_spearman_cosine
0.2778	100	0.0316	0.0274	0.8415	-
0.5556	200	0.0306	0.0280	0.8392	-
0.8333	300	0.0282	0.0280	0.8462	-
1.1111	400	0.0208	0.0277	0.8482	-
1.3889	500	0.0148	0.0271	0.8494	-
1.6667	600	0.0136	0.0259	0.8503	-
1.9444	700	0.0137	0.0259	0.8525	-
2.2222	800	0.0089	0.0262	0.8519	-
2.5	900	0.0074	0.0255	0.8551	-
2.7778	1000	0.0071	0.0256	0.8544	-
3.0556	1100	0.0068	0.0258	0.8558	-
3.3333	1200	0.005	0.0253	0.8565	-
3.6111	1300	0.0046	0.0259	0.8547	-
3.8889	1400	0.0046	0.0257	0.8559	-
4.0	1440	-	-	-	0.8235

Framework Versions

Python: 3.10.12
Sentence Transformers: 3.0.1
Transformers: 4.42.4
PyTorch: 2.3.1+cu121
Accelerate: 0.32.1
Datasets: 2.20.0
Tokenizers: 0.19.1

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",
}