Image-Text-to-Text
Transformers
Safetensors
English
pathology
computational-pathology
digital-pathology
histopathology
whole-slide-image
vision-language-model
report-generation
synoptic-report
case-level
conch
qwen2.5
Eval Results (legacy)
Instructions to use AtlasAnalyticsLab/PathoSynVLM with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use AtlasAnalyticsLab/PathoSynVLM with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="AtlasAnalyticsLab/PathoSynVLM")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("AtlasAnalyticsLab/PathoSynVLM", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use AtlasAnalyticsLab/PathoSynVLM with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "AtlasAnalyticsLab/PathoSynVLM" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AtlasAnalyticsLab/PathoSynVLM", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/AtlasAnalyticsLab/PathoSynVLM
- SGLang
How to use AtlasAnalyticsLab/PathoSynVLM with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "AtlasAnalyticsLab/PathoSynVLM" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AtlasAnalyticsLab/PathoSynVLM", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "AtlasAnalyticsLab/PathoSynVLM" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AtlasAnalyticsLab/PathoSynVLM", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use AtlasAnalyticsLab/PathoSynVLM with Docker Model Runner:
docker model run hf.co/AtlasAnalyticsLab/PathoSynVLM
File size: 7,066 Bytes
aa50ff8 | 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 | <svg xmlns="http://www.w3.org/2000/svg" width="1400" height="680" viewBox="0 0 1400 680" role="img" aria-labelledby="title desc">
<title id="title">PathoSynVLM reported results</title>
<desc id="desc">Stage 1 visual-language alignment and Stage 2 HISTAI case-level report generation metrics. All bars share a zero-to-one scale and exact values appear in separate labels.</desc>
<defs>
<filter id="shadow" x="-8%" y="-8%" width="116%" height="125%">
<feDropShadow dx="0" dy="6" stdDeviation="8" flood-color="#17212b" flood-opacity="0.10"/>
</filter>
<style>
.sans { font-family: Arial, Helvetica, sans-serif; }
.eyebrow { font-size: 14px; font-weight: 700; fill: #0e8a9c; }
.title { font-size: 34px; font-weight: 700; fill: #17212b; }
.subtitle { font-size: 17px; fill: #5c626b; }
.panel-title { font-size: 24px; font-weight: 700; fill: #ffffff; }
.panel-subtitle { font-size: 15px; fill: #d8ecf4; }
.group { font-size: 13px; font-weight: 700; fill: #68727d; }
.metric { font-size: 16px; font-weight: 700; fill: #26313d; }
.value { font-size: 15px; font-weight: 700; fill: #17212b; }
.note { font-size: 14px; fill: #68727d; }
</style>
</defs>
<rect width="1400" height="680" fill="#f7fafc"/>
<g class="sans">
<text x="60" y="45" class="eyebrow">PATHOSYNVLM</text>
<text x="60" y="82" class="title">Reported Results</text>
<text x="60" y="109" class="subtitle">Headline paper metrics. Every bar uses the same 0-1 scale; exact values sit in a dedicated column.</text>
<g filter="url(#shadow)">
<rect x="60" y="140" width="590" height="470" rx="8" fill="#ffffff" stroke="#d6dee7"/>
<path d="M68 140h574a8 8 0 0 1 8 8v82H60v-82a8 8 0 0 1 8-8Z" fill="#0b315b"/>
</g>
<text x="95" y="181" class="panel-title">Stage 1 · VL alignment</text>
<text x="95" y="209" class="panel-subtitle">HistGen + REG2025 · trainable projector only</text>
<text x="95" y="259" class="group">TEXT GENERATION</text>
<g>
<text x="95" y="302" class="metric">ROUGE-L</text>
<rect x="270" y="286" width="240" height="14" rx="7" fill="#e5f3f6"/>
<rect x="270" y="286" width="114" height="14" rx="7" fill="#0e8a9c"/>
<rect x="535" y="275" width="80" height="34" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="575" y="298" text-anchor="middle" class="value">0.4743</text>
<text x="95" y="370" class="metric">METEOR</text>
<rect x="270" y="354" width="240" height="14" rx="7" fill="#e5f3f6"/>
<rect x="270" y="354" width="115" height="14" rx="7" fill="#0e8a9c"/>
<rect x="535" y="343" width="80" height="34" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="575" y="366" text-anchor="middle" class="value">0.4810</text>
<text x="95" y="438" class="metric">BLEU-4</text>
<rect x="270" y="422" width="240" height="14" rx="7" fill="#e5f3f6"/>
<rect x="270" y="422" width="30" height="14" rx="7" fill="#0e8a9c"/>
<rect x="535" y="411" width="80" height="34" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="575" y="434" text-anchor="middle" class="value">0.1247</text>
<text x="95" y="506" class="metric">BERTScore F1</text>
<rect x="270" y="490" width="240" height="14" rx="7" fill="#e5f3f6"/>
<rect x="270" y="490" width="102" height="14" rx="7" fill="#0e8a9c"/>
<rect x="535" y="479" width="80" height="34" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="575" y="502" text-anchor="middle" class="value">0.4253</text>
</g>
<g filter="url(#shadow)">
<rect x="690" y="140" width="650" height="470" rx="8" fill="#ffffff" stroke="#d6dee7"/>
<path d="M698 140h634a8 8 0 0 1 8 8v82H690v-82a8 8 0 0 1 8-8Z" fill="#164f83"/>
</g>
<text x="725" y="181" class="panel-title">Stage 2 · HISTAI</text>
<text x="725" y="209" class="panel-subtitle">Case-level structured report generation</text>
<text x="725" y="254" class="group">TEXT GENERATION</text>
<g>
<text x="725" y="286" class="metric">ROUGE-L</text>
<rect x="960" y="273" width="220" height="12" rx="6" fill="#e5f3f6"/>
<rect x="960" y="273" width="55" height="12" rx="6" fill="#0e8a9c"/>
<rect x="1210" y="262" width="94" height="32" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="1257" y="284" text-anchor="middle" class="value">0.2495</text>
<text x="725" y="326" class="metric">METEOR</text>
<rect x="960" y="313" width="220" height="12" rx="6" fill="#e5f3f6"/>
<rect x="960" y="313" width="44" height="12" rx="6" fill="#0e8a9c"/>
<rect x="1210" y="302" width="94" height="32" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="1257" y="324" text-anchor="middle" class="value">0.1988</text>
<text x="725" y="366" class="metric">BLEU-4</text>
<rect x="960" y="353" width="220" height="12" rx="6" fill="#e5f3f6"/>
<rect x="960" y="353" width="12" height="12" rx="6" fill="#0e8a9c"/>
<rect x="1210" y="342" width="94" height="32" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="1257" y="364" text-anchor="middle" class="value">0.0525</text>
<text x="725" y="406" class="metric">BERTScore F1</text>
<rect x="960" y="393" width="220" height="12" rx="6" fill="#e5f3f6"/>
<rect x="960" y="393" width="66" height="12" rx="6" fill="#0e8a9c"/>
<rect x="1210" y="382" width="94" height="32" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="1257" y="404" text-anchor="middle" class="value">0.3018</text>
</g>
<line x1="725" y1="432" x2="1305" y2="432" stroke="#e0e6ec"/>
<text x="725" y="458" class="group">FIELD ACCURACY</text>
<g>
<text x="725" y="490" class="metric">Diagnosis · exact</text>
<rect x="960" y="477" width="220" height="12" rx="6" fill="#e7f3eb"/>
<rect x="960" y="477" width="37" height="12" rx="6" fill="#3a9e5b"/>
<rect x="1210" y="466" width="94" height="32" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="1257" y="488" text-anchor="middle" class="value">0.1667</text>
<text x="725" y="530" class="metric">Diagnosis · relaxed</text>
<rect x="960" y="517" width="220" height="12" rx="6" fill="#e7f3eb"/>
<rect x="960" y="517" width="73" height="12" rx="6" fill="#3a9e5b"/>
<rect x="1210" y="506" width="94" height="32" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="1257" y="528" text-anchor="middle" class="value">0.3333</text>
<text x="725" y="570" class="metric">Certainty · exact</text>
<rect x="960" y="557" width="220" height="12" rx="6" fill="#e7f3eb"/>
<rect x="960" y="557" width="198" height="12" rx="6" fill="#3a9e5b"/>
<rect x="1210" y="546" width="94" height="32" rx="6" fill="#ffffff" stroke="#bfcbd6"/>
<text x="1257" y="568" text-anchor="middle" class="value">0.9000</text>
</g>
<text x="60" y="648" class="note">BLEU-4 is shown on the paper's 0-1 scale. Exact values are stored in configs/reported_results.json.</text>
</g>
</svg>
|