SentenceTransformer(
(0): Transformer({'max_seq_length': 8192, 'do_lower_case': False, 'architecture': '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})
)pip install -U sentence-transformers1from sentence_transformers import SentenceTransformer
2
3# Download from the 🤗 Hub
4model = SentenceTransformer("sentence_transformers_model_id")
5# Run inference
6sentences = [
7 'the density and density gradient of the=1.99, hi=3.04, hC=1.28, hAr=26.7, hb=0.486, and q background thermal deuterium ion species are adjusted to=2.01. Here, rnj;−sd ln nj/drd−1 and hj satisfy the requirements of charge neutrality and of the radial;sd ln Tj/drd/sd ln nj/drd, and the other notation is stan- derivative of charge neutrality on the chosen magnetic sur-dard. face. The local experimental parameters for the ITG root for There is also a constraint on the output of the calcula-the NSTX ',
8 'the changes in the growth rates and real fre- number of general conclusions. sid The range of density and quencies and the absolute changes in the normalized particle temperature gradient variation surveyed here for the five spe- and energy fluxes are small for all species, sand again would cies often involves regime change, between stable and un- be difficult to see in a figured. However, the relative changes stable, or between ITG-like and TEM-like. siid Usually, when in the normalized beam sp',
9 'means that the power densities can be arranged to be large (many MW m−2), and also one of the ‘figures of merit’ for divertors, B204 A W Morris et al (P/R)divertor, can be arranged to be large and comparable to the ITER values, albeit in a different geometry. Several geometric factors are important in SOL modelling and fitting—for example it is important to know the effect of the increased shaping (κ, δ) proposed for RTO/RC-ITER on the SOL width, and hence the divertor power loading. Scaling law',
10]
11embeddings = model.encode(sentences)
12print(embeddings.shape)
13# [3, 768]
14
15# Get the similarity scores for the embeddings
16similarities = model.similarity(embeddings, embeddings)
17print(similarities)
18# tensor([[1.0000, 0.5986, 0.6121],
19# [0.5986, 1.0000, 0.5670],
20# [0.6121, 0.5670, 1.0000]])sentence_0, sentence_1, and label| sentence_0 | sentence_1 | label | |
|---|---|---|---|
| type | string | string | float |
| details |
|
|
|
| sentence_0 | sentence_1 | label |
|---|---|---|
resistive wall mode | RWM | 1.0 |
phase (after 1.5 s). Using either Nucl. Fusion 58 (2018) 046010 D.J. Battaglia et al version of EFC directly enabled stable, long-pulse L-mode discharges. A final set of preliminary EFC experiments focused on identifying the optimum EFC in the ramp-up phase of L- and H-mode discharges. These experiments found that the optimum early-time n = 1 correction phase was oppo- site to that of the phase in EFC v2. In other words, the best phase of the n = 1 EFC during the flattop of the discharge contrib | had few or no ELMs βN/li that made the discharges more at risk for MHD activity due to the heating power being close to the power required that could lead to a disruption. Measurements of the current to remain in H-mode, thus the impurity content accumulated profile from the motional stark emission (MSE) diagnostic through the H-mode period as inferred from visible imaging were not available during the commissioning phase, thus a and spectroscopy diagnostics. The resulting increase in radi- rigo | 0.6 |
vertical displacement event | VDE | 1.0 |
CosineSimilarityLoss with these parameters:
1{
2 "loss_fct": "torch.nn.modules.loss.MSELoss"
3}per_device_train_batch_size: 4per_device_eval_batch_size: 4multi_dataset_batch_sampler: round_robinoverwrite_output_dir: Falsedo_predict: Falseeval_strategy: noprediction_loss_only: Trueper_device_train_batch_size: 4per_device_eval_batch_size: 4per_gpu_train_batch_size: Noneper_gpu_eval_batch_size: Nonegradient_accumulation_steps: 1eval_accumulation_steps: Nonetorch_empty_cache_steps: Nonelearning_rate: 5e-05weight_decay: 0.0adam_beta1: 0.9adam_beta2: 0.999adam_epsilon: 1e-08max_grad_norm: 1num_train_epochs: 3max_steps: -1lr_scheduler_type: linearlr_scheduler_kwargs: {}warmup_ratio: 0.0warmup_steps: 0log_level: passivelog_level_replica: warninglog_on_each_node: Truelogging_nan_inf_filter: Truesave_safetensors: Truesave_on_each_node: Falsesave_only_model: Falserestore_callback_states_from_checkpoint: Falseno_cuda: Falseuse_cpu: Falseuse_mps_device: Falseseed: 42data_seed: Nonejit_mode_eval: Falsebf16: Falsefp16: Falsefp16_opt_level: O1half_precision_backend: autobf16_full_eval: Falsefp16_full_eval: Falsetf32: Nonelocal_rank: 0ddp_backend: Nonetpu_num_cores: Nonetpu_metrics_debug: Falsedebug: []dataloader_drop_last: Falsedataloader_num_workers: 0dataloader_prefetch_factor: Nonepast_index: -1disable_tqdm: Falseremove_unused_columns: Truelabel_names: Noneload_best_model_at_end: Falseignore_data_skip: Falsefsdp: []fsdp_min_num_params: 0fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap: Noneaccelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}parallelism_config: Nonedeepspeed: Nonelabel_smoothing_factor: 0.0optim: adamw_torch_fusedoptim_args: Noneadafactor: Falsegroup_by_length: Falselength_column_name: lengthproject: huggingfacetrackio_space_id: trackioddp_find_unused_parameters: Noneddp_bucket_cap_mb: Noneddp_broadcast_buffers: Falsedataloader_pin_memory: Truedataloader_persistent_workers: Falseskip_memory_metrics: Trueuse_legacy_prediction_loop: Falsepush_to_hub: Falseresume_from_checkpoint: Nonehub_model_id: Nonehub_strategy: every_savehub_private_repo: Nonehub_always_push: Falsehub_revision: Nonegradient_checkpointing: Falsegradient_checkpointing_kwargs: Noneinclude_inputs_for_metrics: Falseinclude_for_metrics: []eval_do_concat_batches: Truefp16_backend: autopush_to_hub_model_id: Nonepush_to_hub_organization: Nonemp_parameters:auto_find_batch_size: Falsefull_determinism: Falsetorchdynamo: Noneray_scope: lastddp_timeout: 1800torch_compile: Falsetorch_compile_backend: Nonetorch_compile_mode: Noneinclude_tokens_per_second: Falseinclude_num_input_tokens_seen: noneftune_noise_alpha: Noneoptim_target_modules: Nonebatch_eval_metrics: Falseeval_on_start: Falseuse_liger_kernel: Falseliger_kernel_config: Noneeval_use_gather_object: Falseaverage_tokens_across_devices: Trueprompts: Nonebatch_sampler: batch_samplermulti_dataset_batch_sampler: round_robinrouter_mapping: {}learning_rate_mapping: {}1@inproceedings{reimers-2019-sentence-bert,
2 title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
3 author = "Reimers, Nils and Gurevych, Iryna",
4 booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
5 month = "11",
6 year = "2019",
7 publisher = "Association for Computational Linguistics",
8 url = "https://arxiv.org/abs/1908.10084",
9}