{"repo": "avduarte333/sem-detect", "files": {"demo/templates/index.html": {"d97175a513fab1aa97b31cc9a06874bad79a21c0": {"t": 1785747172.8708665, "n": 25193, "h": "0cf1045b15795af4", "c": "\n\n\n \n \n Sem-Detect — AI Review Detection\n \n \n \n \n \n\n\n\n
\n
\n \n
\n \n
\n \n \n
\n
\n
\n
\n\n
\n\n \n
\n\n
\n Analysis in progress. Credentials cannot be changed while a pipeline is running.\n
\n\n
\n

Connect your services

\n

Credentials are stored in-memory for this session only. Nothing is written to disk.

\n
\n\n
\n\n \n
\n
\n
\n \n

Gemini

\n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenAI

\n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

LiteLLM Proxy

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n \n

AWS Bedrock

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

OpenReview

\n \n
\n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n\n \n
\n \n \n
\n\n
\n
\n\n \n
\n\n
\n

Run SEM-DETECT

\n
\n\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n

\n Some conferences allow authors to update blind submissions during the rebuttal period.
\n In these cases, use the cutoff date to ensure we fetch the same version that reviewers saw.\n

\n\n
\n\n \n
\n
\n \n \n
\n
\n
\n \n ▶\n Custom Rating Scale\n \n
\n
\n
\n \n \n
\n
\n \n \n
\n
\n \n
\n
\n \n
\n
\n
\n
\n
\n

\n Maps each numeric score from OpenReview to one of 5 classes used by the pipeline.\n

\n\n
\n\n \n
\n
\n \n

\n Send the original PDF directly to LLMs instead of extracting text with marker-pdf.
\n Faster but won't let you inspect the parsed pdf text. (Some models don't support this feature!)\n

\n
\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n \n \n
\n

We recommend using Gemini-2.5-Pro, DeepSeek and Qwen3.
(Models we used to tune Sem-Detect classifier).

\n
\n\n
\n\n \n
\n
\n \n
Gemini 2.5 Flash
\n \n
\n
\n \n
Qwen3-Embedding-0.6B
\n
\n
\n \n \n

Mistral-7B is our preferred model. Qwen3-1.7B is faster but less accurate.

\n
\n

The pipeline needs local Qwen3-0.6B and the selected textual model.
\n (GPU is recommended for speed)

\n
\n\n
\n\n \n
\n
\n
\n \n
\n
\n\n
\n\n
\n
\n\n \n
\n\n
\n

ICLR 2026 Results

\n

Pre-computed Sem-Detect analysis on 677 ICLR 2026 papers (2,607 reviews).

\n
\n\n
\n
\n \n \n
\n
\n \n \n \n \n
\n
\n\n
\n
\n \n
\n\n \n
\n \n
\n
\n \n
\n\n
\n\n
\n\n \n\n \n \n\n\n"}, "ca5058b454e24875c73fd98799bc6d0417f733f9": {"t": 1785770932.4388564, "n": 25193, "h": "0cf1045b15795af4", "c": "\n\n\n \n \n Sem-Detect — AI Review Detection\n \n \n \n \n \n\n\n\n
\n
\n \n
\n \n
\n \n \n
\n
\n
\n
\n\n
\n\n \n
\n\n
\n Analysis in progress. Credentials cannot be changed while a pipeline is running.\n
\n\n
\n

Connect your services

\n

Credentials are stored in-memory for this session only. Nothing is written to disk.

\n
\n\n
\n\n \n
\n
\n
\n \n

Gemini

\n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenAI

\n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

LiteLLM Proxy

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n \n

AWS Bedrock

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

OpenReview

\n \n
\n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n\n \n
\n \n \n
\n\n
\n
\n\n \n
\n\n
\n

Run SEM-DETECT

\n
\n\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n

\n Some conferences allow authors to update blind submissions during the rebuttal period.
\n In these cases, use the cutoff date to ensure we fetch the same version that reviewers saw.\n

\n\n
\n\n \n
\n
\n \n \n
\n
\n
\n \n ▶\n Custom Rating Scale\n \n
\n
\n
\n \n \n
\n
\n \n \n
\n
\n \n
\n
\n \n
\n
\n
\n
\n
\n

\n Maps each numeric score from OpenReview to one of 5 classes used by the pipeline.\n

\n\n
\n\n \n
\n
\n \n

\n Send the original PDF directly to LLMs instead of extracting text with marker-pdf.
\n Faster but won't let you inspect the parsed pdf text. (Some models don't support this feature!)\n

\n
\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n \n \n
\n

We recommend using Gemini-2.5-Pro, DeepSeek and Qwen3.
(Models we used to tune Sem-Detect classifier).

\n
\n\n
\n\n \n
\n
\n \n
Gemini 2.5 Flash
\n \n
\n
\n \n
Qwen3-Embedding-0.6B
\n
\n
\n \n \n

Mistral-7B is our preferred model. Qwen3-1.7B is faster but less accurate.

\n
\n

The pipeline needs local Qwen3-0.6B and the selected textual model.
\n (GPU is recommended for speed)

\n
\n\n
\n\n \n
\n
\n
\n \n
\n
\n\n
\n\n
\n
\n\n \n
\n\n
\n

ICLR 2026 Results

\n

Pre-computed Sem-Detect analysis on 677 ICLR 2026 papers (2,607 reviews).

\n
\n\n
\n
\n \n \n
\n
\n \n \n \n \n
\n
\n\n
\n
\n \n
\n\n \n
\n \n
\n
\n \n
\n\n
\n\n
\n\n \n\n \n \n\n\n"}, "4bbcbf7f4cfc1ce3dd15dab99dd004ce3bb82312": {"t": 1785770933.9810944, "n": 19071, "h": "0a369457743cc974", "c": "\n\n\n \n \n Sem-Detect — AI Review Detection\n \n \n \n \n \n\n\n\n
\n
\n \n
\n \n
\n \n \n
\n
\n
\n
\n\n
\n\n \n
\n\n
\n Analysis in progress. Credentials cannot be changed while a pipeline is running.\n
\n\n
\n

Connect your services

\n

Credentials are stored in-memory for this session only. Nothing is written to disk.

\n
\n\n
\n\n \n
\n
\n
\n \n

Gemini

\n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenAI

\n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

AWS Bedrock

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenReview

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n \n \n
\n \n
\n\n
\n
\n\n \n
\n\n
\n

Run SEM-DETECT

\n
\n\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n

\n Some conferences allow authors to update blind submissions during the rebuttal period.
\n In these cases, use the cutoff date to ensure we fetch the same version that reviewers saw.\n

\n\n
\n\n \n
\n
\n \n

\n Send the original PDF directly to LLMs instead of extracting text with marker-pdf.
\n Faster but won't let you inspect the parsed pdf text. (Some models don't support this feature!)\n

\n
\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n \n \n
\n

We recommend using Gemini-2.5-Pro, DeepSeek and Qwen3.
(Models we used to tune Sem-Detect classifier).

\n
\n\n
\n\n \n
\n
\n \n
Gemini 2.5 Flash
\n \n
\n
\n \n
Qwen3-Embedding-0.6B
\n
\n
\n \n
Mistral-7B-Instruct-v0.3
\n
\n

The pipeline needs local Qwen3-0.6B and Mistral-7B.
\n (GPU is recommended for speed)

\n
\n\n
\n\n \n
\n
\n
\n \n
\n
\n\n
\n\n
\n
\n\n \n
\n\n
\n

ICLR 2026 Results

\n

Pre-computed Sem-Detect analysis on 677 ICLR 2026 papers (2,607 reviews).

\n
\n\n
\n
\n \n \n
\n
\n \n \n \n \n
\n
\n\n
\n
\n \n
\n\n \n
\n \n
\n
\n \n
\n\n
\n\n
\n\n \n\n \n \n\n\n"}, "e7c2bde13b67c811f35855d7bb9b04004e3e8755": {"t": 1785770935.1776774, "n": 15949, "h": "b5d2e612011d2aa6", "c": "\n\n\n \n \n Sem-Detect — AI Review Detection\n \n\n\n\n
\n
\n \n
\n
\n\n
\n\n \n
\n\n
\n Analysis in progress. Credentials cannot be changed while a pipeline is running.\n
\n\n
\n

Connect your services

\n

Credentials are stored in-memory for this session only. Nothing is written to disk.

\n
\n\n
\n\n \n
\n
\n
\n \n

Gemini

\n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenAI

\n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

AWS Bedrock

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenReview

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n \n \n
\n \n
\n\n
\n
\n\n \n
\n\n
\n

Run SEM-DETECT

\n
\n\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n

\n Some conferences allow authors to update blind submissions during the rebuttal period.
\n In these cases, use the cutoff date to ensure we fetch the same version that reviewers saw.\n

\n\n
\n\n \n
\n
\n \n

\n Send the original PDF directly to LLMs instead of extracting text with marker-pdf.
\n Faster but won't let you inspect the parsed pdf text. (Some models don't support this feature!)\n

\n
\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n \n \n
\n

We recommend using Gemini-2.5-Pro, DeepSeek and Qwen3.
(Models we used to tune Sem-Detect classifier).

\n
\n\n
\n\n \n
\n
\n \n
Gemini 2.5 Flash
\n \n
\n
\n \n
Qwen3-Embedding-0.6B
\n
\n
\n \n
Mistral-7B-Instruct-v0.3
\n
\n

The pipeline needs local Qwen3-0.6B and Mistral-7B.
\n (GPU is recommended for speed)

\n
\n\n
\n\n \n
\n \n
\n
\n \n
\n
\n\n
\n\n
\n
\n\n
\n\n \n\n \n \n\n\n"}, "0ba41decf0616c50b68528d10f73cd716bc57fae": {"t": 1785770936.9752026, "n": 15672, "h": "9460a8af828e7729", "c": "\n\n\n \n \n Sem-Detect — AI Review Detection\n \n\n\n\n
\n
\n \n
\n
\n\n
\n\n \n
\n\n
\n Analysis in progress. Credentials cannot be changed while a pipeline is running.\n
\n\n
\n

Connect your services

\n

Credentials are stored in-memory for this session only. Nothing is written to disk.

\n
\n\n
\n\n \n
\n
\n
\n \n

Gemini

\n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenAI

\n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

AWS Bedrock

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenReview

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n\n \n
\n \n
\n\n
\n
\n\n \n
\n\n
\n

Run SEM-DETECT

\n
\n\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n

\n Some conferences allow authors to update blind submissions during the rebuttal period.
\n In these cases, use the cutoff date to ensure we fetch the same version that reviewers saw.\n

\n\n
\n\n \n
\n
\n \n

\n Send the original PDF directly to LLMs instead of extracting text with marker-pdf.
\n Faster but won't let you inspect the parsed pdf text. (Some models don't support this feature!)\n

\n
\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n \n \n
\n

We recommend using Gemini-2.5-Pro, DeepSeek and Qwen3.
(Models we used to tune Sem-Detect classifier).

\n
\n\n
\n\n \n
\n
\n \n
Gemini 2.5 Flash
\n \n
\n
\n \n
Qwen3-Embedding-0.6B
\n
\n
\n \n
Mistral-7B-Instruct-v0.3
\n
\n

The pipeline needs local Qwen3-0.6B and Mistral-7B.
\n (GPU is recommended for speed)

\n
\n\n
\n\n \n
\n \n
\n
\n \n
\n
\n\n
\n\n
\n
\n\n
\n\n \n\n \n \n\n\n"}, "5db962156f981e1306d466a7f1f4472ca5852755": {"t": 1785770938.4976122, "n": 11096, "h": "77f062105b8268d7", "c": "\n\n\n \n \n Sem-Detect — AI Review Detection\n \n\n\n\n
\n
\n \n
\n
\n\n
\n\n \n
\n\n
\n Analysis in progress. Credentials cannot be changed while a pipeline is running.\n
\n\n
\n

Connect your services

\n

Credentials are stored in-memory for this session only. Nothing is written to disk.

\n
\n\n
\n\n \n
\n
\n
\n \n

Gemini

\n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenAI

\n \n
\n
\n \n \n
\n
\n
\n\n \n
\n
\n
\n \n

AWS Bedrock

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n
\n \n

OpenReview

\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n
\n\n \n
\n \n
\n\n
\n
\n\n \n
\n\n
\n

Run SEM-DETECT

\n

Note: Some conferences allow authors to update blind submissions during the rebuttal period!
Use the cutoff date to ensure we fetch the same version that reviewers saw.

\n
\n\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n\n
\n\n \n
\n
\n \n \n
\n
\n \n \n
\n
\n \n \n
\n
\n \n \n
\n
\n\n
\n\n \n
\n \n
\n
\n \n
\n
\n\n
\n\n
\n
\n\n
\n\n \n\n \n \n\n\n"}}}}