> **Note on embed dim.** Projection head is 320-dim (trained). `config.json` sets `dim: 320` (read by `ColQwen3_5`) with `embed_dim: 320` as a synonym for cross-ecosystem tools, so `ColQwen3_5.from_pretrained(...)` loads correctly by default. Leaderboard columns showing `128` reflect the colpali-engine class default (`self.dim = getattr(config, "dim", 128)`) and not the actual output dim. V3 builds on V2 with automated hyperparameter search and evolutionary model soup merging. The search identified an optimal LoRA config (r=16, alpha=64) with cosine scheduling, and the final checkpoint is a per-layer evolutionary merge with V2. The result is a +0.0219 improvement over V2 across benchmarks. _Rankings on the live ViDoRe V3 leaderboard drift as new submissions land. At release, v3 held **#3 overall**; figures below were captured on 2026-04-20, where v3 sits at **#6 on Mean (Task)** and remains **top-3 among 4B-class models**._

译文排队中 · 当前显示来源原文

分类待补充

输入待核验→ 输出待核验

价格待核验多模态与渠道费用须分别核对

查看详情 ↗

加入对比来源 / 文档 ↗

ColUltraVBERT

athrael-soju

基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。

基础资料已收录 · 缺失资料后续补充

分类待补充

输入待核验→ 输出待核验

价格待核验多模态与渠道费用须分别核对

查看详情 ↗

加入对比来源 / 文档 ↗

llmlingua-2-js-tinybert-meetingbank

atjsh

基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。

分类待补充

输入待核验→ 输出待核验

价格待核验多模态与渠道费用须分别核对

查看详情 ↗

加入对比来源 / 文档 ↗

Selene-1-Llama-3.3-70B

AtlaAI

基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。

基础资料已收录 · 缺失资料后续补充

文本生成

输入文本→ 输出文本

价格待核验多模态与渠道费用须分别核对

查看详情 ↗

加入对比来源 / 文档 ↗

Selene-1-Mini-Llama-3.1-8B

AtlaAI

基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。

文本生成

输入文本→ 输出文本

价格待核验多模态与渠道费用须分别核对

查看详情 ↗

加入对比来源 / 文档 ↗

JEV-27B-VL-GGUF

Atlas3D

GGUF builds of autotrust/JEV-27B-VL for llama.cpp's native decision endpoint `/v1/systemone`, with JEV's own readout: false/true for yes/no, the digits 0–5 for scores, and letters for choices, plus its decision-head bias and calibrated temperatures. These are also the builds for **older GPUs such as V100s**, which cannot run FP8/FP4 checkpoints. > **Requires a patched llama.cpp (for now).** The `jev` decision type is not in upstream llama.cpp yet. > Apply `llama.cpp-jev.patch` (made against upstream `bed0a85`) and build. It adds the converter class, > the GGUF metadata keys and the server readout. > ⚠️ **Q4_K_M is less accurate.** About 1 decision in 14 differs from full precision. Prefer Q8_0 when it > fits. Q4_K_M is for 24 GB-class cards.

译文排队中 · 当前显示来源原文

视觉问答

输入文本 · 图像→ 输出文本

价格待核验多模态与渠道费用须分别核对

查看详情 ↗

加入对比来源 / 文档 ↗

DeepSeek-V4-Flash-0731-FP8-DSpark

AtlasCloud

Routed MoE experts converted MXFP4/FP4 -> block FP8 via lossless `cast_e2m1fn_to_e4m3fn` (`tmp/scripts/convert_hf_mxfp4_experts_to_fp8.py`). `expert_dtype` removed from `config.json`; other fields (including `dspark_*` if present) kept.

译文排队中 · 当前显示来源原文

分类待补充

输入待核验→ 输出待核验

价格待核验多模态与渠道费用须分别核对

查看详情 ↗

加入对比来源 / 文档 ↗