This is a sentence-transformers model finetuned from Snowflake/snowflake-arctic-embed-l. It maps sentences & paragraphs to a 1024-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: Snowflake/snowflake-arctic-embed-l
Maximum Sequence Length: 512 tokens
Output Dimensionality: 1024 dimensions
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': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Normalize()
)

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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("CalebMaresca/matrix-game-embeddings-ft-v1")
# Run inference
sentences = [
    ' Why is it important for player roles in a Matrix Game to operate at broadly similar levels?',
    'When you are designing a Matrix Game it is worth thinking about the level at which the players roles will be operating in the game. In is usually better, and produces a more balanced game, when the level on which the player roles are operating are broadly similar. It would be difficult to get a balanced game if 3 of the players are playing Generals in command of vast Armies, and another player is playing a simple individual soldier.\n\nLevels of Protection and Hidden Things.',
    'Matrix Game Checklist .................................................................................... 38 \nSample Spendable Bonus Cards ...................................................................... 40 \nSample Random Events ................................................................................... 41 \nSample Voting Cards for Diceless Adjudication ............................................... 43 \nSample Estimative Probability Cards ............................................................... 44 \nSample Turn Order Cards ................................................................................ 45 \nSample Markers for Matrix Games for Effects and Conventional Forces ........ 46',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

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

Evaluation

Metrics

Information Retrieval

Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.9348
cosine_accuracy@3	1.0
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	0.9348
cosine_precision@3	0.3333
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	0.9348
cosine_recall@3	1.0
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	0.9702
cosine_mrr@10	0.9601
cosine_map@100	0.9601

Training Details

Training Dataset

Unnamed Dataset

Size: 370 training samples
Columns: sentence_0 and sentence_1
Approximate statistics based on the first 370 samples:
sentence_0 sentence_1
type string string
details
min: 11 tokens
mean: 20.19 tokens
max: 34 tokens
min: 8 tokens
mean: 150.83 tokens
max: 512 tokens

Samples:

sentence_0	sentence_1
`What distinguishes "established facts" from other types of facts in the game briefings or play?`	Forces soldiers are going to be much more effective in combat than untrained protestors; [object Object]and "established facts" which are facts that have been specifically mentioned in the game [object Object]briefings or have become established during play as the result of successful arguments. [object Object]The latter can be immediately deployed as supporting reasons (Pros and Cons), but the [object Object]former need to have been argued successfully in order for them to be specifically included. [object Object]Many inexperienced players will make vast all-encompassing arguments full of assumptions [object Object]that are not reasonable. For example: It is not a reasonable assumption that unarmed [object Object]Protestors could fight off trained Police. It is reasonable to assume that the Police are
`Why is it unreasonable to assume that unarmed protestors could fight off trained police according to the context?`	Forces soldiers are going to be much more effective in combat than untrained protestors; [object Object]and "established facts" which are facts that have been specifically mentioned in the game [object Object]briefings or have become established during play as the result of successful arguments. [object Object]The latter can be immediately deployed as supporting reasons (Pros and Cons), but the [object Object]former need to have been argued successfully in order for them to be specifically included. [object Object]Many inexperienced players will make vast all-encompassing arguments full of assumptions [object Object]that are not reasonable. For example: It is not a reasonable assumption that unarmed [object Object]Protestors could fight off trained Police. It is reasonable to assume that the Police are
`What was the outcome of the initial Russian attack against the German units, and how did it affect the ammunition status of both sides?`	The Russians succeed in pushing back one of the German units and forcing and already depleted unit to use up ammunition, (but are pushed back themselves and 2 units use a lot of ammo (one of which becomes combat ineffective on -3)). Overall, as the success is matched by failure, the line itself holds. The Russians attack again, the next day:[object Object][object Object]Initial Dice Throw: RUSSIAN: 6 5 5 4 2 4 GERMAN: 1 2 4 4 Lined Up and Modified: RUSSIAN: 5 4 3 3 2 1 (two of the Russians = -2) GERMAN: 4 3 3 2 (one of the Germans = -1) Result of Third Day; lose: (one of the Germans = +0) RUSSIAN: 5 4 3 3 GERMAN: 3 lose: 4 3 2

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 768, 5 512, 6 256, 7 128, 8 64 9 ], 10 "matryoshka_weights": [ 11 1, 12 1, 13 1, 14 1, 15 1 16 ], 17 "n_dims_per_step": -1 18}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 10
per_device_eval_batch_size: 10
num_train_epochs: 10
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 10
per_device_eval_batch_size: 10
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: 5e-05
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1
num_train_epochs: 10
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
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: 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}
tp_size: 0
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
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
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: round_robin

Training Logs

Epoch	Step	cosine_ndcg@10
1.0	37	0.9273
1.3514	50	0.9490
2.0	74	0.9462
2.7027	100	0.9527
3.0	111	0.9527
4.0	148	0.9783
4.0541	150	0.9811
5.0	185	0.9622
5.4054	200	0.9622
6.0	222	0.9702
6.7568	250	0.9622
7.0	259	0.9622
8.0	296	0.9702
8.1081	300	0.9702
9.0	333	0.9702
9.4595	350	0.9702
10.0	370	0.9702

Framework Versions

Python: 3.13.2
Sentence Transformers: 4.1.0
Transformers: 4.51.3
PyTorch: 2.7.0+cu126
Accelerate: 1.6.0
Datasets: 3.6.0
Tokenizers: 0.21.1

Citation

BibTeX

Sentence Transformers

bibtex

@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",
}

MatryoshkaLoss

bibtex

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

MultipleNegativesRankingLoss

bibtex

@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}
}

	sentence_0	sentence_1
type	string	string
details	min: 11 tokens mean: 20.19 tokens max: 34 tokens	min: 8 tokens mean: 150.83 tokens max: 512 tokens