pearmut 0.3.3__tar.gz → 1.0.0__tar.gz

{pearmut-0.3.3 → pearmut-1.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pearmut
-Version: 0.3.3
+Version: 1.0.0
 Summary: A tool for evaluation of model outputs, primarily MT.
 Author-email: Vilém Zouhar <vilem.zouhar@gmail.com>
 License: MIT
@@ -14,6 +14,7 @@ Requires-Dist: fastapi>=0.110.0
 Requires-Dist: uvicorn>=0.29.0
 Requires-Dist: wonderwords>=3.0.0
 Requires-Dist: psutil>=7.1.0
+Requires-Dist: typst>=0.14.4
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Dynamic: license-file
@@ -30,7 +31,8 @@ Dynamic: license-file
 &nbsp;
 [![build status](https://github.com/zouharvi/pearmut/actions/workflows/test.yml/badge.svg)](https://github.com/zouharvi/pearmut/actions/workflows/test.yml)
 
-<img width="1000" alt="Screenshot of ESA/MQM interface" src="https://github.com/user-attachments/assets/4fb9a1cb-78ac-47e0-99cd-0870a368a0ad" />
+<img width="1000" alt="Screenshot of ESA/MQM interface" src="https://github.com/user-attachments/assets/71334238-300b-4ffc-b777-7f3c242b1630" />
+
 
 ## Table of Contents
 
@@ -45,6 +47,7 @@ Dynamic: license-file
   - [Multimodal Annotations](#multimodal-annotations)
   - [Hosting Assets](#hosting-assets)
 - [Campaign Management](#campaign-management)
+  - [Custom Completion Messages](#custom-completion-messages)
 - [CLI Commands](#cli-commands)
 - [Terminology](#terminology)
 - [Development](#development)
@@ -86,11 +89,13 @@ Campaigns are defined in JSON files (see [examples/](examples/)). The simplest c
         {
           "instructions": "Evaluate translation from en to cs_CZ",  # message to show to users above the first item
           "src": "This will be the year that Guinness loses its cool. Cheers to that!",
-          "tgt": {"modelA": "Nevím přesně, kdy jsem to poprvé zaznamenal. Možná to bylo ve chvíli, ..."}
+          "tgt": {"modelA": "Nevím přesně, kdy jsem to poprvé zaznamenal. Možná to bylo ve chvíli, ..."},
+          "item_id": "first item in first document"
         },
         {
           "src": "I'm not sure I can remember exactly when I sensed it. Maybe it was when some...",
-          "tgt": {"modelA": "Tohle bude rok, kdy Guinness přijde o svůj „cool“ faktor. Na zdraví!"}
+          "tgt": {"modelA": "Tohle bude rok, kdy Guinness přijde o svůj „cool“ faktor. Na zdraví!"},
+          "item_id": "second item in first document"
         }
         ...
       ],
@@ -105,20 +110,10 @@ Campaigns are defined in JSON files (see [examples/](examples/)). The simplest c
   ]
 }
 ```
-Task items are protocol-specific. For ESA/DA/MQM protocols, each item is a dictionary representing a document unit:
-```python
-[
-  {
-    "src": "A najednou se všechna tato voda naplnila dalšími lidmi a dalšími věcmi.",  # required
-    "tgt": {"modelA": "And suddenly all the water became full of other people and other people."}  # required (dict)
-  },
-  {
-    "src": "toto je pokračování stejného dokumentu",
-    "tgt": {"modelA": "this is a continuation of the same document"}
-    # Additional keys stored for analysis
-  }
-]
-```
+
+Each item has to have `src` (string) and `tgt` (dictionary from model names to strings, even for a single model evaluation).
+For full Pearmut functionality (e.g. automatic statistical analysis), add `item_id` as well.
+Any other keys that you add will simply be stored in the logs.
 
 Load campaigns and start the server:
 ```bash
@@ -130,7 +125,7 @@ pearmut run
 
 - **`task-based`**: Each user has predefined items
 - **`single-stream`**: All users draw from a shared pool (random assignment)
-- **`dynamic`**: work in progress ⚠️
+- **`dynamic`**: Items are dynamically assigned based on current model performance (see [Dynamic Assignment](#dynamic-assignment))
 
 ## Advanced Features
 
@@ -244,6 +239,36 @@ All annotators draw from a shared pool with random assignment:
 }
 ```
 
+### Dynamic Assignment
+
+The `dynamic` assignment type intelligently selects items based on current model performance to focus annotation effort on top-performing models using contrastive comparisons.
+All items must contain outputs from all models for this assignment type to work properly.
+
+```python
+{
+    "campaign_id": "my dynamic campaign",
+    "info": {
+        "assignment": "dynamic",
+        "protocol": "ESA",
+        "users": 10,                           # number of annotators
+        "dynamic_top": 3,                      # how many top models to consider (required)
+        "dynamic_contrastive_models": 2,       # how many models to compare per item (optional, default: 1)
+        "dynamic_first": 5,                    # annotations per model before dynamic kicks in (optional, default: 5)
+        "dynamic_backoff": 0.1,                # probability of uniform sampling (optional, default: 0)
+    },
+    "data": [...], # list of all items (shared among all annotators)
+}
+```
+
+**How it works:**
+1. Initial phase: Each model gets `dynamic_first` annotations with fully random contrastive evaluation
+2. Dynamic phase: After the initial phase, top `dynamic_top` models (by average score) are identified
+3. Contrastive evaluatoin: From the top N models, `dynamic_contrastive_models` models are randomly selected for each item
+4. Item prioritization: Items with the least annotations for the selected models are prioritized
+5. Backoff: With probability `dynamic_backoff`, uniform random selection is used instead to maintain exploration
+
+This approach efficiently focuses annotation resources on distinguishing between the best-performing models while ensuring all models get adequate baseline coverage. The contrastive evaluation allows for direct comparison of multiple models simultaneously.
+For an example, see [examples/dynamic.json](examples/dynamic.json).
 
 ### Pre-defined User IDs and Tokens
 
@@ -317,6 +342,10 @@ Completion tokens are shown at annotation end for verification (download correct
 
 When tokens are supplied, the dashboard will try to show model rankings based on the names in the dictionaries.
 
+### Custom Completion Messages
+
+Customize the goodbye message shown to users when they complete all annotations using the `instructions_goodbye` field in campaign info. Supports arbitrary HTML for styling and formatting with variable replacement: `${TOKEN}` (completion token) and `${USER_ID}` (user ID). Default: `"If someone asks you for a token of completion, show them: ${TOKEN}"`.
+
 ## Terminology
 
 - **Campaign**: An annotation project that contains configuration, data, and user assignments. Each campaign has a unique identifier and is defined in a JSON file.
@@ -344,7 +373,7 @@ When tokens are supplied, the dashboard will try to show model rankings based on
 - **Assignment**: The method for distributing items to users:
   - **Task-based**: Each user has predefined items
   - **Single-stream**: Users draw from a shared pool with random assignment
-  - **Dynamic**: Work in progress
+  - **Dynamic**: Items are intelligently assigned based on model performance to focus on top models
 
 ## Development
 
@@ -377,15 +406,14 @@ See [web/src/basic.ts](web/src/basic.ts) for example.
 
 Run on public server or tunnel local port to public IP/domain and run locally.
 
-## Misc.
+## Citation
 
 If you use this work in your paper, please cite as following.
 ```bibtex
-@misc{zouhar2025pearmut,
-    author={Vilém Zouhar},
-    title={Pearmut: Platform for Evaluating and Reviewing of Multilingual Tasks},
-    url={https://github.com/zouharvi/pearmut/},
-    year={2026},
+@misc{zouhar2026pearmut,
+  author = {Zouhar, Vilém},
+  title = {Pearmut: Human Evaluation of Translation Made Trivial},
+  year = {2026}
 }
 ```
 

{pearmut-0.3.3 → pearmut-1.0.0}/README.md

@@ -10,7 +10,8 @@
  
 [![build status](https://github.com/zouharvi/pearmut/actions/workflows/test.yml/badge.svg)](https://github.com/zouharvi/pearmut/actions/workflows/test.yml)
 
-<img width="1000" alt="Screenshot of ESA/MQM interface" src="https://github.com/user-attachments/assets/4fb9a1cb-78ac-47e0-99cd-0870a368a0ad" />
+<img width="1000" alt="Screenshot of ESA/MQM interface" src="https://github.com/user-attachments/assets/71334238-300b-4ffc-b777-7f3c242b1630" />
+
 
 ## Table of Contents
 
@@ -25,6 +26,7 @@
   - [Multimodal Annotations](#multimodal-annotations)
   - [Hosting Assets](#hosting-assets)
 - [Campaign Management](#campaign-management)
+  - [Custom Completion Messages](#custom-completion-messages)
 - [CLI Commands](#cli-commands)
 - [Terminology](#terminology)
 - [Development](#development)
@@ -66,11 +68,13 @@ Campaigns are defined in JSON files (see [examples/](examples/)). The simplest c
         {
           "instructions": "Evaluate translation from en to cs_CZ",  # message to show to users above the first item
           "src": "This will be the year that Guinness loses its cool. Cheers to that!",
-          "tgt": {"modelA": "Nevím přesně, kdy jsem to poprvé zaznamenal. Možná to bylo ve chvíli, ..."}
+          "tgt": {"modelA": "Nevím přesně, kdy jsem to poprvé zaznamenal. Možná to bylo ve chvíli, ..."},
+          "item_id": "first item in first document"
         },
         {
           "src": "I'm not sure I can remember exactly when I sensed it. Maybe it was when some...",
-          "tgt": {"modelA": "Tohle bude rok, kdy Guinness přijde o svůj „cool“ faktor. Na zdraví!"}
+          "tgt": {"modelA": "Tohle bude rok, kdy Guinness přijde o svůj „cool“ faktor. Na zdraví!"},
+          "item_id": "second item in first document"
         }
         ...
       ],
@@ -85,20 +89,10 @@ Campaigns are defined in JSON files (see [examples/](examples/)). The simplest c
   ]
 }
 ```
-Task items are protocol-specific. For ESA/DA/MQM protocols, each item is a dictionary representing a document unit:
-```python
-[
-  {
-    "src": "A najednou se všechna tato voda naplnila dalšími lidmi a dalšími věcmi.",  # required
-    "tgt": {"modelA": "And suddenly all the water became full of other people and other people."}  # required (dict)
-  },
-  {
-    "src": "toto je pokračování stejného dokumentu",
-    "tgt": {"modelA": "this is a continuation of the same document"}
-    # Additional keys stored for analysis
-  }
-]
-```
+
+Each item has to have `src` (string) and `tgt` (dictionary from model names to strings, even for a single model evaluation).
+For full Pearmut functionality (e.g. automatic statistical analysis), add `item_id` as well.
+Any other keys that you add will simply be stored in the logs.
 
 Load campaigns and start the server:
 ```bash
@@ -110,7 +104,7 @@ pearmut run
 
 - **`task-based`**: Each user has predefined items
 - **`single-stream`**: All users draw from a shared pool (random assignment)
-- **`dynamic`**: work in progress ⚠️
+- **`dynamic`**: Items are dynamically assigned based on current model performance (see [Dynamic Assignment](#dynamic-assignment))
 
 ## Advanced Features
 
@@ -224,6 +218,36 @@ All annotators draw from a shared pool with random assignment:
 }
 ```
 
+### Dynamic Assignment
+
+The `dynamic` assignment type intelligently selects items based on current model performance to focus annotation effort on top-performing models using contrastive comparisons.
+All items must contain outputs from all models for this assignment type to work properly.
+
+```python
+{
+    "campaign_id": "my dynamic campaign",
+    "info": {
+        "assignment": "dynamic",
+        "protocol": "ESA",
+        "users": 10,                           # number of annotators
+        "dynamic_top": 3,                      # how many top models to consider (required)
+        "dynamic_contrastive_models": 2,       # how many models to compare per item (optional, default: 1)
+        "dynamic_first": 5,                    # annotations per model before dynamic kicks in (optional, default: 5)
+        "dynamic_backoff": 0.1,                # probability of uniform sampling (optional, default: 0)
+    },
+    "data": [...], # list of all items (shared among all annotators)
+}
+```
+
+**How it works:**
+1. Initial phase: Each model gets `dynamic_first` annotations with fully random contrastive evaluation
+2. Dynamic phase: After the initial phase, top `dynamic_top` models (by average score) are identified
+3. Contrastive evaluatoin: From the top N models, `dynamic_contrastive_models` models are randomly selected for each item
+4. Item prioritization: Items with the least annotations for the selected models are prioritized
+5. Backoff: With probability `dynamic_backoff`, uniform random selection is used instead to maintain exploration
+
+This approach efficiently focuses annotation resources on distinguishing between the best-performing models while ensuring all models get adequate baseline coverage. The contrastive evaluation allows for direct comparison of multiple models simultaneously.
+For an example, see [examples/dynamic.json](examples/dynamic.json).
 
 ### Pre-defined User IDs and Tokens
 
@@ -297,6 +321,10 @@ Completion tokens are shown at annotation end for verification (download correct
 
 When tokens are supplied, the dashboard will try to show model rankings based on the names in the dictionaries.
 
+### Custom Completion Messages
+
+Customize the goodbye message shown to users when they complete all annotations using the `instructions_goodbye` field in campaign info. Supports arbitrary HTML for styling and formatting with variable replacement: `${TOKEN}` (completion token) and `${USER_ID}` (user ID). Default: `"If someone asks you for a token of completion, show them: ${TOKEN}"`.
+
 ## Terminology
 
 - **Campaign**: An annotation project that contains configuration, data, and user assignments. Each campaign has a unique identifier and is defined in a JSON file.
@@ -324,7 +352,7 @@ When tokens are supplied, the dashboard will try to show model rankings based on
 - **Assignment**: The method for distributing items to users:
   - **Task-based**: Each user has predefined items
   - **Single-stream**: Users draw from a shared pool with random assignment
-  - **Dynamic**: Work in progress
+  - **Dynamic**: Items are intelligently assigned based on model performance to focus on top models
 
 ## Development
 
@@ -357,15 +385,14 @@ See [web/src/basic.ts](web/src/basic.ts) for example.
 
 Run on public server or tunnel local port to public IP/domain and run locally.
 
-## Misc.
+## Citation
 
 If you use this work in your paper, please cite as following.
 ```bibtex
-@misc{zouhar2025pearmut,
-    author={Vilém Zouhar},
-    title={Pearmut: Platform for Evaluating and Reviewing of Multilingual Tasks},
-    url={https://github.com/zouharvi/pearmut/},
-    year={2026},
+@misc{zouhar2026pearmut,
+  author = {Zouhar, Vilém},
+  title = {Pearmut: Human Evaluation of Translation Made Trivial},
+  year = {2026}
 }
 ```
 

{pearmut-0.3.3 → pearmut-1.0.0}/pearmut.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pearmut
-Version: 0.3.3
+Version: 1.0.0
 Summary: A tool for evaluation of model outputs, primarily MT.
 Author-email: Vilém Zouhar <vilem.zouhar@gmail.com>
 License: MIT
@@ -14,6 +14,7 @@ Requires-Dist: fastapi>=0.110.0
 Requires-Dist: uvicorn>=0.29.0
 Requires-Dist: wonderwords>=3.0.0
 Requires-Dist: psutil>=7.1.0
+Requires-Dist: typst>=0.14.4
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Dynamic: license-file
@@ -30,7 +31,8 @@ Dynamic: license-file
 &nbsp;
 [![build status](https://github.com/zouharvi/pearmut/actions/workflows/test.yml/badge.svg)](https://github.com/zouharvi/pearmut/actions/workflows/test.yml)
 
-<img width="1000" alt="Screenshot of ESA/MQM interface" src="https://github.com/user-attachments/assets/4fb9a1cb-78ac-47e0-99cd-0870a368a0ad" />
+<img width="1000" alt="Screenshot of ESA/MQM interface" src="https://github.com/user-attachments/assets/71334238-300b-4ffc-b777-7f3c242b1630" />
+
 
 ## Table of Contents
 
@@ -45,6 +47,7 @@ Dynamic: license-file
   - [Multimodal Annotations](#multimodal-annotations)
   - [Hosting Assets](#hosting-assets)
 - [Campaign Management](#campaign-management)
+  - [Custom Completion Messages](#custom-completion-messages)
 - [CLI Commands](#cli-commands)
 - [Terminology](#terminology)
 - [Development](#development)
@@ -86,11 +89,13 @@ Campaigns are defined in JSON files (see [examples/](examples/)). The simplest c
         {
           "instructions": "Evaluate translation from en to cs_CZ",  # message to show to users above the first item
           "src": "This will be the year that Guinness loses its cool. Cheers to that!",
-          "tgt": {"modelA": "Nevím přesně, kdy jsem to poprvé zaznamenal. Možná to bylo ve chvíli, ..."}
+          "tgt": {"modelA": "Nevím přesně, kdy jsem to poprvé zaznamenal. Možná to bylo ve chvíli, ..."},
+          "item_id": "first item in first document"
         },
         {
           "src": "I'm not sure I can remember exactly when I sensed it. Maybe it was when some...",
-          "tgt": {"modelA": "Tohle bude rok, kdy Guinness přijde o svůj „cool“ faktor. Na zdraví!"}
+          "tgt": {"modelA": "Tohle bude rok, kdy Guinness přijde o svůj „cool“ faktor. Na zdraví!"},
+          "item_id": "second item in first document"
         }
         ...
       ],
@@ -105,20 +110,10 @@ Campaigns are defined in JSON files (see [examples/](examples/)). The simplest c
   ]
 }
 ```
-Task items are protocol-specific. For ESA/DA/MQM protocols, each item is a dictionary representing a document unit:
-```python
-[
-  {
-    "src": "A najednou se všechna tato voda naplnila dalšími lidmi a dalšími věcmi.",  # required
-    "tgt": {"modelA": "And suddenly all the water became full of other people and other people."}  # required (dict)
-  },
-  {
-    "src": "toto je pokračování stejného dokumentu",
-    "tgt": {"modelA": "this is a continuation of the same document"}
-    # Additional keys stored for analysis
-  }
-]
-```
+
+Each item has to have `src` (string) and `tgt` (dictionary from model names to strings, even for a single model evaluation).
+For full Pearmut functionality (e.g. automatic statistical analysis), add `item_id` as well.
+Any other keys that you add will simply be stored in the logs.
 
 Load campaigns and start the server:
 ```bash
@@ -130,7 +125,7 @@ pearmut run
 
 - **`task-based`**: Each user has predefined items
 - **`single-stream`**: All users draw from a shared pool (random assignment)
-- **`dynamic`**: work in progress ⚠️
+- **`dynamic`**: Items are dynamically assigned based on current model performance (see [Dynamic Assignment](#dynamic-assignment))
 
 ## Advanced Features
 
@@ -244,6 +239,36 @@ All annotators draw from a shared pool with random assignment:
 }
 ```
 
+### Dynamic Assignment
+
+The `dynamic` assignment type intelligently selects items based on current model performance to focus annotation effort on top-performing models using contrastive comparisons.
+All items must contain outputs from all models for this assignment type to work properly.
+
+```python
+{
+    "campaign_id": "my dynamic campaign",
+    "info": {
+        "assignment": "dynamic",
+        "protocol": "ESA",
+        "users": 10,                           # number of annotators
+        "dynamic_top": 3,                      # how many top models to consider (required)
+        "dynamic_contrastive_models": 2,       # how many models to compare per item (optional, default: 1)
+        "dynamic_first": 5,                    # annotations per model before dynamic kicks in (optional, default: 5)
+        "dynamic_backoff": 0.1,                # probability of uniform sampling (optional, default: 0)
+    },
+    "data": [...], # list of all items (shared among all annotators)
+}
+```
+
+**How it works:**
+1. Initial phase: Each model gets `dynamic_first` annotations with fully random contrastive evaluation
+2. Dynamic phase: After the initial phase, top `dynamic_top` models (by average score) are identified
+3. Contrastive evaluatoin: From the top N models, `dynamic_contrastive_models` models are randomly selected for each item
+4. Item prioritization: Items with the least annotations for the selected models are prioritized
+5. Backoff: With probability `dynamic_backoff`, uniform random selection is used instead to maintain exploration
+
+This approach efficiently focuses annotation resources on distinguishing between the best-performing models while ensuring all models get adequate baseline coverage. The contrastive evaluation allows for direct comparison of multiple models simultaneously.
+For an example, see [examples/dynamic.json](examples/dynamic.json).
 
 ### Pre-defined User IDs and Tokens
 
@@ -317,6 +342,10 @@ Completion tokens are shown at annotation end for verification (download correct
 
 When tokens are supplied, the dashboard will try to show model rankings based on the names in the dictionaries.
 
+### Custom Completion Messages
+
+Customize the goodbye message shown to users when they complete all annotations using the `instructions_goodbye` field in campaign info. Supports arbitrary HTML for styling and formatting with variable replacement: `${TOKEN}` (completion token) and `${USER_ID}` (user ID). Default: `"If someone asks you for a token of completion, show them: ${TOKEN}"`.
+
 ## Terminology
 
 - **Campaign**: An annotation project that contains configuration, data, and user assignments. Each campaign has a unique identifier and is defined in a JSON file.
@@ -344,7 +373,7 @@ When tokens are supplied, the dashboard will try to show model rankings based on
 - **Assignment**: The method for distributing items to users:
   - **Task-based**: Each user has predefined items
   - **Single-stream**: Users draw from a shared pool with random assignment
-  - **Dynamic**: Work in progress
+  - **Dynamic**: Items are intelligently assigned based on model performance to focus on top models
 
 ## Development
 
@@ -377,15 +406,14 @@ See [web/src/basic.ts](web/src/basic.ts) for example.
 
 Run on public server or tunnel local port to public IP/domain and run locally.
 
-## Misc.
+## Citation
 
 If you use this work in your paper, please cite as following.
 ```bibtex
-@misc{zouhar2025pearmut,
-    author={Vilém Zouhar},
-    title={Pearmut: Platform for Evaluating and Reviewing of Multilingual Tasks},
-    url={https://github.com/zouharvi/pearmut/},
-    year={2026},
+@misc{zouhar2026pearmut,
+  author = {Zouhar, Vilém},
+  title = {Pearmut: Human Evaluation of Translation Made Trivial},
+  year = {2026}
 }
 ```
 

{pearmut-0.3.3 → pearmut-1.0.0}/pearmut.egg-info/SOURCES.txt

@@ -10,6 +10,7 @@ pearmut.egg-info/top_level.txt
 server/app.py
 server/assignment.py
 server/cli.py
+server/results_export.py
 server/utils.py
 server/static/basic.bundle.js
 server/static/basic.html

{pearmut-0.3.3 → pearmut-1.0.0}/pearmut.egg-info/requires.txt

@@ -2,6 +2,7 @@ fastapi>=0.110.0
 uvicorn>=0.29.0
 wonderwords>=3.0.0
 psutil>=7.1.0
+typst>=0.14.4
 
 [dev]
 pytest

{pearmut-0.3.3 → pearmut-1.0.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "pearmut"
-version = "0.3.3"
+version = "1.0.0"
 description = "A tool for evaluation of model outputs, primarily MT."
 readme = "README.md"
 license = { text = "MIT" }
@@ -17,6 +17,7 @@ dependencies = [
     "uvicorn >= 0.29.0",
     "wonderwords >= 3.0.0",
     "psutil >= 7.1.0",
+    "typst >= 0.14.4",
 ]
 
 [project.optional-dependencies]

{pearmut-0.3.3 → pearmut-1.0.0}/server/app.py

@@ -1,20 +1,23 @@
-import collections
 import json
 import os
-import statistics
 from typing import Any
 
 from fastapi import FastAPI, Query
 from fastapi.middleware.cors import CORSMiddleware
-from fastapi.responses import JSONResponse
+from fastapi.responses import JSONResponse, Response
 from fastapi.staticfiles import StaticFiles
 from pydantic import BaseModel
 
 from .assignment import get_i_item, get_next_item, reset_task, update_progress
+from .results_export import (
+    compute_model_scores,
+    generate_latex_table,
+    generate_pdf,
+    generate_typst_table,
+)
 from .utils import (
     ROOT,
     check_validation_threshold,
-    get_db_log,
     load_progress_data,
     save_db_payload,
     save_progress_data,
@@ -159,7 +162,7 @@ async def _dashboard_data(request: DashboardDataRequest):
 
     progress_new = {}
     assignment = tasks_data[campaign_id]["info"]["assignment"]
-    if assignment not in ["task-based", "single-stream"]:
+    if assignment not in ["task-based", "single-stream", "dynamic"]:
         return JSONResponse(
             content="Unsupported campaign assignment type", status_code=400
         )
@@ -211,31 +214,47 @@ async def _dashboard_results(request: DashboardResultsRequest):
     if token != tasks_data[campaign_id]["token"]:
         return JSONResponse(content="Invalid token", status_code=400)
 
-    # Compute model scores from annotations
-    model_scores = collections.defaultdict(dict)
-
-    # Iterate through all tasks to find items with 'models' field (basic template)
-    log = get_db_log(campaign_id)
-    for entry in log:
-        if "item" not in entry or "annotation" not in entry:
-            continue
-        for item, annotation in zip(entry["item"], entry["annotation"]):
-            for model, annotation in annotation.items():
-                if "score" in annotation and annotation["score"] is not None:
-                    model_scores[model][json.dumps(item)] = annotation["score"]
-
-    results = [
-        {
-            "model": model,
-            "score": statistics.mean(scores.values()),
-            "count": len(scores),
-        }
-        for model, scores in model_scores.items()
-    ]
-    results.sort(key=lambda x: x["score"], reverse=True)
+    results = compute_model_scores(campaign_id)
     return JSONResponse(content=results, status_code=200)
 
 
+@app.get("/export-results")
+async def _export_results(
+    campaign_id: str = Query(),
+    token: str = Query(),
+    format: str = Query(),
+):
+    if campaign_id not in progress_data:
+        return JSONResponse(content="Unknown campaign ID", status_code=400)
+
+    # Check if token is valid
+    if token != tasks_data[campaign_id]["token"]:
+        return JSONResponse(content="Invalid token", status_code=400)
+
+    results = compute_model_scores(campaign_id)
+
+    if format == "typst":
+        content = generate_typst_table(results)
+        return Response(
+            content=content,
+            media_type="text/plain",
+        )
+    elif format == "latex":
+        content = generate_latex_table(results)
+        return Response(
+            content=content,
+            media_type="text/plain",
+        )
+    elif format == "pdf":
+        pdf_bytes = generate_pdf(results, campaign_id)
+        return Response(
+            content=pdf_bytes,
+            media_type="application/pdf",
+        )
+    else:
+        return JSONResponse(content="Invalid export format", status_code=400)
+
+
 class ResetTaskRequest(BaseModel):
     campaign_id: str
     user_id: str