Spaces:
Running
Running
File size: 32,997 Bytes
6ba6082 564134f 6ba6082 564134f 6ba6082 564134f 6ba6082 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 |
<!DOCTYPE html>
<html data-theme="light">
<head>
<meta charset="utf-8" />
<meta name="description" content="Building an Open Polish Vision-Language Model." />
<meta name="keywords" content="VLM, Polish, AI, Multimodal, LLM, PLLuM, LLaVA-PLLuM" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<title>LLaVA-PLLuM: a Polish Vision-Language Model</title>
<link rel="icon"
href="data:image/svg+xml,<svg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 36 36'><text x='0' y='32' font-size='32'>🇵🇱</text></svg>" />
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro" rel="stylesheet" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bulma@1.0.4/css/bulma.min.css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bulma-carousel@4.0.3/dist/css/bulma-carousel.min.css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bulma-slider@2.0.4/dist/css/bulma-slider.min.css" />
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/5.15.3/css/all.min.css" />
<link rel="stylesheet" href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css" />
<link rel="stylesheet" href="./static/css/index.css" />
<link rel="stylesheet" href="./static/css/custom.css" />
<link rel="icon" href="./static/images/favicon.svg" />
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script src="https://cdn.jsdelivr.net/npm/bulma-carousel@4.0.3/dist/js/bulma-carousel.min.js"></script>
<script src="https://cdn.jsdelivr.net/npm/bulma-slider@2.0.4/dist/js/bulma-slider.min.js"></script>
<script src="https://unpkg.com/lucide@latest"></script>
<script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
<script src="./static/js/index.js"></script>
</head>
<body>
<section class="hero">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">🇵🇱 LLaVA-PLLuM: Building an Open Polish
Vision-Language Model</h1>
<h2 class="subtitle is-4 publication-subtitle mt-4">
Bridging the gap in multilingual AI with culturally-aware image understanding
</h2>
<div class="is-size-5 publication-authors">
<div class="publication-authors">
<strong>Grzegorz Statkiewicz, Alicja Dobrzeniecka, Aleksandra Krasnodębska, Sebastian Cygert, Wojciech
Kusa</strong>
</div>
<div class="author-institute">
<span class="author-block">NASK National Research Institute</span>
</div>
<span class="author-block" style="margin-left: 15px">
<a href="mailto:firstname.lastname@nask.pl">firstname.lastname@nask.pl</a>
</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a class="external-link button is-normal is-rounded is-dark" disabled>
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span>arXiv (soon)</span>
</a>
</span>
<!-- Code Link. -->
<span class="link-block">
<a href="https://huggingface.co/NASK-PIB/LLaVA-PLLuM-12B-nc-instruct"
target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i data-lucide="download"></i>
</span>
<span>Model</span>
</a>
</span>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="section">
<div class="container is-max-desktop">
<div class="columns">
<!-- Table of Contents -->
<div class="column is-3 is-hidden-touch">
<aside class="menu sticky-menu">
<p class="menu-label">Contents</p>
<ul class="menu-list">
<li><a href="#introduction">Introduction</a></li>
<li><a href="#methodology">Methodology</a></li>
<li><a href="#evaluation">Evaluation & Results</a></li>
<li><a href="#qualitative">Qualitative Results</a></li>
<li><a href="#summary">Summary</a></li>
<li><a href="#references">References</a></li>
<li><a href="#bibtex">BibTeX</a></li>
</ul>
</aside>
</div>
<!-- Main Content -->
<div class="column is-9">
<!-- Introduction. -->
<div class="columns is-centered" id="introduction">
<div class="column is-full-width">
<h2 class="title is-3">Introduction</h2>
<div class="content has-text-justified">
<p>
Recent advances in multimodal large language models (MLLMs) have shown impressive capabilities in
combining text and visual understanding. However, most state-of-the-art solutions are trained
primarily on English data, which limits their applicability in other languages and cultural
contexts. Our goal is to bridge this gap by creating a Polish multimodal model that not only
understands text and images but also reflects Polish linguistic and cultural nuances.
</p>
<p>
In this blog post, we describe the methodology used to deliver a proof-of-concept for a Polish Large
Language Model capable of handling both text and visual data. Our approach builds on the LLaVA-NeXT
framework <a href="#ref-3">[3]</a>, which aligns a pretrained visual encoder with a large language
model (LLM) via a lightweight MLP (Multi-Layer Perceptron) projector. We use the following
components:
</p>
<ul>
<li>
<strong>Language Model:</strong> PLLuM-12B (Polish Large Language Model)
<a href="#ref-1">[1]</a> - a Polish-native, instruction-tuned LLM.
</li>
<li>
<strong>Vision Encoder:</strong> SigLIP2 So400m/14, 384px <a href="#ref-4">[4]</a> - chosen for
strong multilingual image-text alignment and improved localization.
</li>
</ul>
<p>
We trained our models using automatic translation combined with manual filtering, resulting in
approximately 550 thousand samples for pretraining and 2 million samples for instruction fine-tuning.
The models
accurately describe images, incorporate Polish cultural context, and handle basic visual tasks such as
OCR and object counting.
</p>
<p>
Evaluation on open-source benchmarks and qualitative analysis show notable improvements in Polish
language understanding as well as recognizing some of Polish cultural elements, while maintaining
general image understanding and reasoning capabilities compared to existing open-source models.
</p>
<p>
This proof-of-concept marks an initial step toward robust multimodal models for Polish. To accelerate
progress and foster collaboration, we are releasing our model weights on Hugging Face.
</p>
</div>
</div>
</div>
<!--/ Introduction. -->
<hr class="section-divider" />
<!-- Methodology -->
<div class="columns is-centered" id="methodology">
<div class="column is-full-width">
<h2 class="title is-3">Methodology</h2>
<h3 class="title is-4">Model Architecture</h3>
<div class="content has-text-justified">
<p>
We build on the LLaVA-NeXT architecture <a href="#ref-3">[3]</a> which aligns a pretrained visual
encoder with a large language model (LLM) via a lightweight two-layer MLP projector. This design
preserves the LLM’s strong language prior while enabling efficient multimodal grounding. Compared to
the original LLaVA, LLaVA-NeXT supports higher input resolutions and dynamic tiling, features that
have been observed to improve fine-grained perception and OCR performance.
</p>
<p>
As the language backbone, we use <strong>PLLuM-12B-nc-instruct-250715</strong>
<a href="#ref-1">[1]</a>, a Polish-native, instruction-tuned LLM. For the vision tower, we replace
the CLIP-like encoder commonly used in LLaVA variants with
<strong>SigLIP2 So400m/14, 384px</strong> <a href="#ref-4">[4]</a>, selected for its strong
multilingual image-text alignment.
</p>
</div>
<h3 class="title is-4">Training Procedure</h3>
<div class="content has-text-justified">
<p>We train the model in two stages following the LLaVA-NeXT procedure:</p>
<ul>
<li>
<strong>Stage 1 (Pre-training):</strong> Freeze the LLM backbone and vision encoder, optimize only
the MLP projector on pretraining dataset to align the connector.
</li>
<li>
<strong>Stage 2 (Instruction Tuning):</strong> Joint training of vision tower and projector,
with LLM adaptation using LoRA on instruction dataset.
</li>
</ul>
<div class="table-container">
<table class="table is-bordered is-striped is-hoverable is-fullwidth">
<thead>
<tr>
<th>Parameter</th>
<th>Stage 1</th>
<th>Stage 2</th>
</tr>
</thead>
<tbody>
<tr>
<td><strong>Training Samples</strong></td>
<td>558K</td>
<td>2M</td>
</tr>
<tr>
<td><strong>Vision Encoder (Trainable)</strong></td>
<td>N/A</td>
<td>400M</td>
</tr>
<tr>
<td><strong>Projector (Trainable)</strong></td>
<td>30M</td>
<td>30M</td>
</tr>
<tr>
<td><strong>Language Model (Trainable)</strong></td>
<td>N/A</td>
<td>12B</td>
</tr>
<tr>
<td><strong>Context Size (Tokens)</strong></td>
<td>8,192</td>
<td>8,192</td>
</tr>
<tr>
<td><strong>Batch Size</strong></td>
<td>256</td>
<td>128</td>
</tr>
<tr>
<td><strong>Learning Rate (Vision Encoder)</strong></td>
<td>N/A</td>
<td>2x10⁻⁶</td>
</tr>
<tr>
<td><strong>Learning Rate (Projector)</strong></td>
<td>1x10⁻³</td>
<td>2x10⁻⁵</td>
</tr>
<tr>
<td><strong>Learning Rate (Language Model)</strong></td>
<td>N/A</td>
<td>2x10⁻⁵</td>
</tr>
<tr>
<td><strong>LoRA Rank (Language Model)</strong></td>
<td>N/A</td>
<td>128</td>
</tr>
<tr>
<td><strong>LoRA Alpha (Language Model)</strong></td>
<td>N/A</td>
<td>256</td>
</tr>
<tr>
<td><strong>LoRA Dropout (Language Model)</strong></td>
<td>N/A</td>
<td>0.05</td>
</tr>
<tr>
<td><strong>Epochs</strong></td>
<td>1</td>
<td>1</td>
</tr>
</tbody>
</table>
</div>
</div>
<h3 class="title is-4">Training Data</h3>
<div class="content has-text-justified">
<p>As the pretraining dataset, we use the LLaVA-LCS-558K <a href="#ref-16">[16]</a> following the LLaVA
paper <a href="#ref-19">[19]</a>. This dataset is a subset of the LAION/CC/SBU collection, filtered
for balanced concept coverage. It consists of 558k image-caption pairs augmented with BLIP synthetic
captions, which we translate to Polish to align the visual features with our language model.</p>
<p>Our instruction dataset spans four skill categories:</p>
<ul>
<li><strong>General:</strong> We translate English datasets: ALLaVA <a href="#ref-5">[5]</a>,
LLaVA-Instruct <a href="#ref-6">[6]</a>, Q-Instruct <a href="#ref-7">[7]</a>, LVIS-Instruct4V <a
href="#ref-8">[8]</a>, and A-OKVQA <a href="#ref-9">[9]</a>.</li>
<li>
<strong>OCR:</strong> Synthetic document-style images. We generate a Polish version (SynthDoG-PL)
and use the English version (SynthDoG-EN) following the SynthDoG procedure <a
href="#ref-10">[10]</a>.
</li>
<li>
<strong>Knowledge:</strong> Based on the WIT dataset <a href="#ref-12">[12]</a>. We select samples
with human-written Polish and English captions.
</li>
<li><strong>Counting:</strong> We translate TallyQA <a href="#ref-13">[13]</a></li>
</ul>
<p>
For translation, we use the Tower+ 72B model <a href="#ref-14">[14]</a> and the COMET reference-free
metric <a href="#ref-15">[15]</a> for filtering.
The resulting datasets are instantiated mostly in Polish (85%), with a smaller sample in English
(15%).
</p>
<div class="column is-full">
<div class="box is-shadowless">
<canvas id="trainingDataChart"></canvas>
<p class="has-text-centered mt-5 has-text-grey">Fine-tuning data distribution</p>
</div>
</div>
</div>
</div>
</div>
<!--/ Methodology -->
<hr class="section-divider" />
<!-- Evaluation -->
<div class="columns is-centered" id="evaluation">
<div class="column is-full-width">
<h2 class="title is-3">Evaluation & Results</h2>
We conduct a two-fold evaluation to assess the performance of our Polish vision-language model: (1)
quantitative benchmarking using MMBench v1.1, and (2) a model-as-a-judge study on image captioning quality
in Polish.
<h3 class="title is-4">MMBench v1.1</h3>
<div class="content has-text-justified">
<p>
Due to the absence of established multimodal evaluation benchmarks in Polish, we adopt existing
English benchmarks for quantitative assessment.
As a primary benchmark, we selected MMBench v1.1 <a href="#ref-17">[17]</a>, which evaluates multiple
dimensions of visual understanding, including object recognition, OCR, commonsense reasoning, and
fine-grained perception.
Because the official MMBench test split has not been released, we choose to evaluate on the
development set.
</p>
<p>
To enable Polish evaluation, we translated all MMBench v1.1 questions into Polish using Tower+ 72B <a
href="#ref-14">[14]</a>, followed by manual expert correction to ensure linguistic accuracy and
eliminate translation artifacts. The resulting MMBench-PL dataset is therefore human-validated and
suitable for assessing Polish multimodal reasoning.
</p>
<p>
The usage of development split makes comparisons strictly fair only against the LLaVA family of
models, whose training data and fine-tuning procedures are publicly documented. For other open-source
VLMs (e.g., Pixtral, Qwen2.5-VL, PaliGemma), the extent of exposure to MMBench during fine-tuning is
unknown.
Only PaliGemma partially discloses pre-training information, but not fine-tuning, and therefore direct
leaderboard-style comparison should be interpreted with caution.
</p>
<div class="table-container">
<table class="table is-bordered is-striped is-hoverable is-fullwidth">
<thead>
<tr>
<th>Model</th>
<th>MMBench (Polish)</th>
<th>MMBench (English)</th>
</tr>
</thead>
<tbody>
<tr>
<td>LLaVA-1.6-Mistral-7B</td>
<td>66.41%</td>
<td>72.37%</td>
</tr>
<tr>
<td>LLaVA-1.6-Vicuna-13B</td>
<td>68.29%</td>
<td>74.14%</td>
</tr>
<tr class="is-selected">
<td><strong>LLaVA-PLLuM-12b-nc (Ours)</strong></td>
<td><strong>73.89%</strong> <span class="tag is-success">+5.6%</span></td>
<td><strong>73.89%</strong></td>
</tr>
<tr class="has-background-light">
<td colspan="3" class="has-text-centered">
<em>Additional Open-Source Models (different architectures)</em>
</td>
</tr>
<tr>
<td>PaliGemma2-10B</td>
<td>77.63%</td>
<td>79.59%</td>
</tr>
<tr>
<td>Pixtral-12B</td>
<td>79.04%</td>
<td>81.52%</td>
</tr>
<tr>
<td>Qwen2.5-VL-7B</td>
<td>74.38%</td>
<td>79.02%</td>
</tr>
</tbody>
</table>
</div>
<p>
<strong>Key Finding:</strong> Our model achieves <strong>+5.6% improvement</strong> on Polish
benchmark compared to LLaVA-1.6-Vicuna-13B while maintaining comparable English performance,
demonstrating significantly improved recognition of Polish context.
</p>
</div>
<h3 class="title is-4">Model-as-a-Judge Evaluation</h3>
<div class="content has-text-justified">
<p>
To evaluate abilities that go beyond multiple-choice recognition and involve open-ended text
generation, we conducted a second study based on image captioning. For this purpose, we used the
Polish portion of the XM3600 dataset [<a href="#ref-18">18</a>].
The task in XM3600 requires models to produce accurate, relevant, and grammatically correct
descriptions of images, making it a suitable testbed for generative multimodal performance.
</p>
<p>
We benchmarked our model against three competitive open-source vision-language models of different
architectures: Qwen2.5-VL-7B-Instruct, Pixtral-12B, and PaliGemma-3B, complementing the MMBench
evaluation.
</p>
<p>
Because no Polish human-annotated standard for caption quality currently exists, we adopted an
LLM-as-a-judge evaluation strategy using LLaVA-OneVision-72B, the strongest open-source VLM at the
time of evaluation and capable of jointly processing the image and candidate captions.
We used a pairwise comparison setup in which the judge is presented with an image and two captions and
determines which description is better.
Since prompt wording and input order can influence the outcome, we employed two prompt
formulations—one presenting caption A before B and one reversing the order—and tested each with both
model assignments (our model as A and as B).
The resulting four judgments for each comparison were then averaged to obtain a stable final score.
</p>
<p>
Together, these steps provide a controlled and replicable protocol for assessing Polish-language
caption quality in the absence of human-annotated ground truth, while capturing the generative
multimodal capabilities of the evaluated models.
</p>
<div class="table-container">
<table class="table is-bordered is-striped is-hoverable is-fullwidth">
<thead>
<tr>
<th>Comparison</th>
<th>Vision-Language Model Judge Winrate</th>
</tr>
</thead>
<tbody>
<tr>
<td>LLaVA-PLLuM-12b-nc vs PaliGemma-3B</td>
<td><strong>95.2%</strong> vs 4.8%</td>
</tr>
<tr>
<td>LLaVA-PLLuM-12b-nc vs Qwen2.5-VL-7B</td>
<td><strong>62.7%</strong> vs 37.3%</td>
</tr>
<tr>
<td>LLaVA-PLLuM-12b-nc vs Pixtral-12B</td>
<td><strong>59.3%</strong> vs 40.7%</td>
</tr>
</tbody>
</table>
</div>
<p>
<strong>Key Finding:</strong> Across all comparisons, LLaVA-PLLuM is consistently preferred by the judge,
indicating higher caption quality in Polish. Our qualitative analysis showed that LLaVA-PLLuM produces more
grammatically correct sentences, maintains proper Polish morphology, and avoids inventing non-existent
Polish words—a common failure mode observed in baseline models.
</p>
</div>
</div>
</div>
<!--/ Evaluation -->
<hr class="section-divider" />
<!-- Qualitative Results -->
<div class="columns is-centered" id="qualitative">
<div class="column is-full-width">
<h2 class="title is-3">Qualitative Results</h2>
<div class="content has-text-justified">
<p>
To examine the models’ ability to grasp and understand Polish cultural context, we collected and
annotated a small dataset of pictures.
</p>
<div id="qualitative-results-container"></div>
</div>
</div>
</div>
<!--/ Qualitative Results -->
<hr class="section-divider" />
<!-- Summary -->
<div class="columns is-centered" id="summary">
<div class="column is-full-width">
<h2 class="title is-3">Summary & Next Steps</h2>
<div class="content has-text-justified">
<p>
We have presented our pipeline for creating: a Polish vision-language model.
Crucially, this system was developed with minimal data curation, relying primarily on synthetic and
machine-translated datasets, without human correction or manual
annotation. Starting from the open-source LLaVA model family and equipping it with the PLLuM language
model, we managed to improve the VLM's ability to understand the Polish language as well as aspects of
Polish cultural context. We show gains of 5.6 percentage points over LLaVA-based baselines on a
manually corrected Polish-language version of MMBench dataset, underscoring the effectiveness of our data-efficient
approach.
</p>
<p>
This is only the first step toward creating a more capable family of Polish vision-language models.
We expect that further scaling of data and leveraging more recent vision-language architectures will
lead to additional improvements. We also intend to enhance the evaluation protocols by incorporating
human assessments and expanding the benchmark datasets to better capture Polish-specific challenges.
</p>
</div>
</div>
</div>
<!--/ Summary -->
<hr class="section-divider" />
<!-- References -->
<div class="columns is-centered" id="references">
<div class="column is-full-width">
<h2 class="title is-3">References</h2>
<div class="content">
<ol>
<li id="ref-1">
PLLuM: A Family of Polish Large Language Models -
<a href="https://arxiv.org/abs/2511.03823">
arXiv:2511.03823
</a>
</li>
<li id="ref-2">
PLLuM Model -
<a href="https://huggingface.co/CYFRAGOVPL/pllum-12b-nc-instruct-250715">
Hugging Face
</a>
</li>
<li id="ref-3">
LLaVA-NeXT -
<a href="https://llava-vl.github.io/blog/2024-01-30-llava-next/">
Blog Post
</a>
</li>
<li id="ref-4">
SigLIP2 -
<a href="https://arxiv.org/abs/2502.14786">
arXiv:2502.14786
</a>
</li>
<li id="ref-5">
ALLaVA -
<a href="https://arxiv.org/abs/2402.11684">
arXiv:2402.11684
</a>
</li>
<li id="ref-6">
Visual Instruction Tuning (LLaVA) -
<a href="https://arxiv.org/abs/2304.08485">
arXiv:2304.08485
</a>
</li>
<li id="ref-7">
Q-Instruct -
<a href="https://arxiv.org/abs/2311.06783">
arXiv:2311.06783
</a>
</li>
<li id="ref-8">
LVIS-Instruct4V -
<a href="https://arxiv.org/abs/2311.07574">
arXiv:2311.07574
</a>
</li>
<li id="ref-9">
A-OKVQA -
<a href="https://arxiv.org/abs/2206.01718">
arXiv:2206.01718
</a>
</li>
<li id="ref-10">
SynthDoG -
<a href="https://arxiv.org/abs/2111.15664">
arXiv:2111.15664
</a>
</li>
<li id="ref-11">
MS COCO -
<a href="https://arxiv.org/abs/1405.0312">
arXiv:1405.0312
</a>
</li>
<li id="ref-12">
WIT Dataset -
<a href="https://doi.org/10.1145/3404835.3463257">
ACM Digital Library
</a>
</li>
<li id="ref-13">
TallyQA -
<a href="https://arxiv.org/abs/1810.12440">
arXiv:1810.12440
</a>
</li>
<li id="ref-14">
Tower+ Translation Model -
<a href="https://huggingface.co/Unbabel/Tower-Plus-72B">
Hugging Face
</a>
</li>
<li id="ref-15">
COMET Metric -
<a href="https://unbabel.github.io/COMET/html/index.html">
Documentation
</a>
</li>
<li id="ref-16">
LLaVA-Pretrain Dataset -
<a href="https://huggingface.co/datasets/liuhaotian/LLaVA-Pretrain">
Hugging Face
</a>
</li>
<li id="ref-17">
MMBench -
<a href="https://huggingface.co/spaces/opencompass/open_vlm_leaderboard">
OpenCompass Leaderboard
</a>
</li>
<li id="ref-18">
Crossmodal-3600: A Massively Multilingual Multimodal Evaluation Dataset -
<a href="https://aclanthology.org/2022.emnlp-main.45/">
EMNLP 2022
</a>
</li>
<li id="ref-19">
Improved Baselines with Visual Instruction Tuning (LLaVA-1.5) -
<a href="https://arxiv.org/abs/2310.03744">
arXiv:2310.03744
</a>
</ol>
</div>
</div>
</div>
<!-- BibTeX -->
<div class="columns is-centered" id="bibtex">
<div class="column is-full-width">
<h2 class="title is-3">BibTeX</h2>
<pre><code>
@misc{statkiewicz2025llavapllum,
title={LLaVA-PLLuM: Building an Open Polish Vision-Language Model},
author={Statkiewicz, Grzegorz and Dobrzeniecka, Alicja and
Krasnodębska, Aleksandra and Cygert, Sebastian and Kusa, Wojciech},
year={2025},
note={Blog post}
}
</code></pre>
</div>
</div>
</div>
</div>
</div>
</section>
<footer class="footer">
<div class="container">
<div class="content has-text-centered">
<p>
This website is adapted from <a href="https://github.com/nerfies/nerfies.github.io" target="_blank">Nerfies</a>, licensed
under a
<a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike
4.0 International License</a>.
</p>
</div>
</div>
</footer>
<button id="scrollToTopBtn" class="button is-rounded is-dark" title="Go to top">
<span class="icon">
<i class="fas fa-arrow-up"></i>
</span>
</button>
</body>
</html> |