diff --git a/data/autora/prompts/all_prompt.json b/data/autora/prompts/all_prompt.json
new file mode 100644
index 0000000..af19692
--- /dev/null
+++ b/data/autora/prompts/all_prompt.json
@@ -0,0 +1,14 @@
+[
+  {
+    "SYS": "You are a technical documentation writer. You always write clear, concise, and accurate documentation for\nscientific experiments. Your documentation focuses on the experiment's purpose, procedure, and results. Therefore,\ndetails about specific python functions, packages, or libraries are not necessary. Your readers are experimental\nscientists.",
+    "INSTR": "Please generate high-level one or two paragraph documentation for the following experiment."
+  },
+  {
+    "SYS": "You are a technical documentation writer. You always write clear, concise, and accurate documentation\nfor scientific experiments. Your documentation focuses on the experiment's procedure. Therefore, details about specific\npython functions, packages, or libraries are NOT necessary. Your readers are experimental scientists.\nFor writing your descriptions, follow these instructions:\n- DO NOT write greetings or preambles\n- Use the Variable 'name' attribute and not the python variable names\n- Use LaTeX for math expressions\n- DO NOT include code or code-like syntax and do not use python function or class names\n- Write in paragraph style, NOT bullet points",
+    "INSTR": "Generate a one line description of the dependent and independent variables used in the following\npython code: "
+  },
+  {
+    "SYS": "You are a research scientist. You always write clear, concise, and accurate documentation\nfor scientific experiments from python code. Your documentation focuses on the experiment's procedure. Therefore, details about specific\npython functions, packages, or libraries are NOT necessary. Your readers are experimental scientists.\nFor writing your descriptions, follow these instructions:\n- DO NOT write greetings or preambles\n- Use the Variable 'name' attribute and not the python variable names\n- Use LaTeX for math expressions\n- DO NOT include code or code-like syntax and do not use python function or class names\n- Write in paragraph style, NOT bullet points",
+    "INSTR": "Generate a three line description of the dependent and independent variables used in the following\npython code: "
+  }
+]
diff --git a/src/autora/doc/classes/EvalResult.py b/src/autora/doc/classes/EvalResult.py
new file mode 100644
index 0000000..993a848
--- /dev/null
+++ b/src/autora/doc/classes/EvalResult.py
@@ -0,0 +1,12 @@
+from dataclasses import dataclass
+from typing import List, Optional
+
+
+@dataclass
+class EvalResult:
+    """Class for storing LLM evaluation results"""
+
+    prediction: List[str]
+    prompt: str
+    bleu_score: Optional[float] = None
+    meteor_score: Optional[float] = None
diff --git a/src/autora/doc/pipelines/main.py b/src/autora/doc/pipelines/main.py
index 2c122c7..33d6504 100644
--- a/src/autora/doc/pipelines/main.py
+++ b/src/autora/doc/pipelines/main.py
@@ -9,8 +9,10 @@
 from nltk.translate.bleu_score import SmoothingFunction, corpus_bleu
 from nltk.translate.meteor_score import single_meteor_score
 
+from autora.doc.classes.EvalResult import EvalResult
 from autora.doc.runtime.predict_hf import Predictor
 from autora.doc.runtime.prompts import PROMPTS, PromptIds
+from autora.doc.util import get_prompts_from_file
 
 app = typer.Typer()
 logging.basicConfig(
@@ -47,6 +49,50 @@ def evaluate_documentation(predictions: List[str], references: List[str]) -> Tup
     return (bleu, meteor)
 
 
+@app.command(help="Evaluate a model for code-to-documentation generation for all prompts in the prompts_file")
+def eval_prompts(
+    data_file: str = typer.Argument(..., help="JSONL Data file to evaluate on"),
+    model_path: str = typer.Option("meta-llama/Llama-2-7b-chat-hf", help="Path to HF model"),
+    prompts_file: str = typer.Argument(..., help="JSON file with a list of dictionary of prompts"),
+    param: List[str] = typer.Option(
+        [], help="Additional float parameters to pass to the model as name=float pairs"
+    ),
+) -> List[EvalResult]:
+    import mlflow
+
+    results_list = []
+
+    mlflow.autolog()
+    param_dict = {pair[0]: float(pair[1]) for pair in [pair.split("=") for pair in param]}
+    run = mlflow.active_run()
+
+    prompts_list = get_prompts_from_file(prompts_file)
+
+    if run is None:
+        run = mlflow.start_run()
+    with run:
+        logger.info(f"Active run_id: {run.info.run_id}")
+        logger.info(f"running predict with {data_file}")
+        logger.info(f"model path: {model_path}")
+        mlflow.log_params(param_dict)
+        mlflow.log_param("model_path", model_path)
+        mlflow.log_param("data_file", data_file)
+        mlflow.log_param("prompts_file", prompts_file)
+        predictor = Predictor(model_path)
+        for i in range(len(prompts_list)):
+            logger.info(f"Starting to run model on prompt {i}")
+            prediction_with_scores = eval_prompt(data_file, predictor, prompts_list[i], param_dict)
+            logger.info(f"Model run completed on prompt {i}: {prompts_list[i]}")
+            eval_result = EvalResult(
+                prediction_with_scores[0],
+                prompts_list[i],
+                prediction_with_scores[1],
+                prediction_with_scores[2],
+            )
+            results_list.append(eval_result)
+        return results_list
+
+
 @app.command(help="Evaluate model on a data file")
 def eval(
     data_file: str = typer.Argument(..., help="JSONL Data file to evaluate on"),
diff --git a/src/autora/doc/util.py b/src/autora/doc/util.py
new file mode 100644
index 0000000..6a63fa0
--- /dev/null
+++ b/src/autora/doc/util.py
@@ -0,0 +1,17 @@
+import json
+from typing import Any, Dict, List, Tuple
+
+from autora.doc.runtime.prompts import PromptBuilder
+
+
+def load_file(json_file_path: str) -> List[Dict[str, Any]]:
+    # Read and parse the JSON file
+    with open(json_file_path, "r") as file:
+        data: List[Dict[str, Any]] = json.load(file)
+    return data
+
+
+def get_prompts_from_file(prompts_file: str) -> List[str]:
+    prompts_data = load_file(prompts_file)
+    prompts_list = [PromptBuilder(p["SYS"], p["INSTR"]).build() for p in prompts_data]
+    return prompts_list
diff --git a/tests/test_main.py b/tests/test_main.py
index 46a74f5..2437be1 100644
--- a/tests/test_main.py
+++ b/tests/test_main.py
@@ -1,9 +1,11 @@
 from pathlib import Path
+from typing import Dict, List
 
 import jsonlines
 import pytest
 
-from autora.doc.pipelines.main import eval, evaluate_documentation, generate, import_data
+from autora.doc.classes.EvalResult import EvalResult
+from autora.doc.pipelines.main import eval, eval_prompts, evaluate_documentation, generate, import_data
 from autora.doc.runtime.prompts import PromptIds
 
 # dummy HF model for testing
@@ -84,3 +86,13 @@ def test_import(tmp_path: Path) -> None:
     import_data(str(code), str(text), str(data))
     new_lines = data.read_text().splitlines()
     assert len(new_lines) == 1, "Expected one new line"
+
+
+def test_eval_prompts() -> None:
+    data_file = Path(__file__).parent.joinpath("../data/sweetpea/data.jsonl").resolve()
+    prompts_file = Path(__file__).parent.joinpath("../data/autora/prompts/all_prompt.json").resolve()
+    results: List[EvalResult] = eval_prompts(str(data_file), TEST_HF_MODEL, str(prompts_file), [])
+    assert len(results) == 3, "Expected 3 outputs"
+    for result in results:
+        assert result.prediction is not None, "The prediction should not be None"
+        assert result.prompt is not None, "The prompt should not be None"
diff --git a/tests/test_util.py b/tests/test_util.py
new file mode 100644
index 0000000..42d6db2
--- /dev/null
+++ b/tests/test_util.py
@@ -0,0 +1,18 @@
+from pathlib import Path
+
+from autora.doc.util import get_prompts_from_file, load_file
+
+
+def test_load_file() -> None:
+    prompts_file_path = Path(__file__).parent.joinpath("../data/autora/prompts/all_prompt.json").resolve()
+    data = load_file(str(prompts_file_path))
+    assert type(data) == list
+
+
+def test_get_prompts_from_file() -> None:
+    prompts_file_path = Path(__file__).parent.joinpath("../data/autora/prompts/all_prompt.json").resolve()
+    prompts_list = get_prompts_from_file(str(prompts_file_path))
+
+    assert len(prompts_list) == 3, "Expected 3 outputs"
+    for prompt in prompts_list:
+        assert type(prompt) == str