Commit
·
eb096cf
1
Parent(s):
b7c29eb
Added tests module and reference runs
Browse files
.gitignore
CHANGED
|
@@ -177,4 +177,4 @@ cython_debug/
|
|
| 177 |
.gradio
|
| 178 |
|
| 179 |
# Test runs
|
| 180 |
-
|
|
|
|
| 177 |
.gradio
|
| 178 |
|
| 179 |
# Test runs
|
| 180 |
+
results_test_case_*.json
|
src/tests.py
ADDED
|
@@ -0,0 +1,187 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import shutil
|
| 2 |
+
import json
|
| 3 |
+
import os
|
| 4 |
+
import time
|
| 5 |
+
import itertools
|
| 6 |
+
import datasets
|
| 7 |
+
import faiss
|
| 8 |
+
import models
|
| 9 |
+
import indexes
|
| 10 |
+
import commons
|
| 11 |
+
from customlogger import log_time, logger
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
def build_field_selection_maps(fields: list[str]) -> dict:
|
| 15 |
+
"""Build all combinations of fields for proverb selection."""
|
| 16 |
+
combos = []
|
| 17 |
+
for r in range(1, len(fields) + 1):
|
| 18 |
+
combos.extend(itertools.combinations(fields, r))
|
| 19 |
+
|
| 20 |
+
maps = {}
|
| 21 |
+
for combo in combos:
|
| 22 |
+
maps["_".join(combo)] = (
|
| 23 |
+
lambda proverb, combo=combo:
|
| 24 |
+
[proverb[field] for field in combo if field != "themes"]
|
| 25 |
+
# Treat "themes" field differently, since it is an array
|
| 26 |
+
+ (proverb["themes"] if "themes" in combo else [])
|
| 27 |
+
)
|
| 28 |
+
return maps
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
def setup():
|
| 32 |
+
"""Set up the environment by loading the model, tokenizer, and dataset."""
|
| 33 |
+
# Load tokenizer and model
|
| 34 |
+
tokenizer = models.load_tokenizer()
|
| 35 |
+
model = models.load_model()
|
| 36 |
+
|
| 37 |
+
# Load proverbs dataset
|
| 38 |
+
proverbs = datasets.load_proverbs()
|
| 39 |
+
prompts = datasets.load_prompts()
|
| 40 |
+
|
| 41 |
+
# By default, the train ratio is zero,
|
| 42 |
+
# but we might still want to do some training in the future
|
| 43 |
+
if datasets.prompts_dataset_splits_exists():
|
| 44 |
+
# Load existing prompt dataset splits
|
| 45 |
+
_, prompts_test_set = datasets.load_prompt_dataset_splits()
|
| 46 |
+
else:
|
| 47 |
+
# Split the prompt dataset into train and test sets
|
| 48 |
+
_, prompts_test_set = datasets.split_dataset(prompts)
|
| 49 |
+
|
| 50 |
+
return tokenizer, model, proverbs, prompts_test_set
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
@log_time
|
| 54 |
+
def test_distances(tokenizer: models.Tokenizer, model: models.Tokenizer, model_name: str,
|
| 55 |
+
proverbs: list[dict], prompts_test_set: list[dict],
|
| 56 |
+
map: tuple[str, callable], index_type: type, pooling_method: str,
|
| 57 |
+
remarks: str = "") -> dict:
|
| 58 |
+
"""Test the distances between the actual and expected proverbs."""
|
| 59 |
+
# Create an index of the type specified from the proverbs dataset with the given map
|
| 60 |
+
embeddings = commons.embed_dataset(
|
| 61 |
+
proverbs, tokenizer, model, map=map[1], pooling_method=pooling_method)
|
| 62 |
+
index = indexes.create_index(embeddings, index_type)
|
| 63 |
+
|
| 64 |
+
# Perform inference on the test prompts
|
| 65 |
+
test_prompts = [entry["prompt"] for entry in prompts_test_set]
|
| 66 |
+
results = commons.inference(
|
| 67 |
+
test_prompts, index, tokenizer, model, proverbs, pooling_method)
|
| 68 |
+
actual_proverbs_embeddings = [result["embedding"] for result in results]
|
| 69 |
+
|
| 70 |
+
# Build a mapping from proverb text to its index for efficient lookup
|
| 71 |
+
proverb_to_index = {proverb["proverb"]
|
| 72 |
+
: i for i, proverb in enumerate(proverbs)}
|
| 73 |
+
|
| 74 |
+
# Find each test proverb in the proverbs dataset and recover its embedding
|
| 75 |
+
test_proverbs = [entry["proverb"] for entry in prompts_test_set]
|
| 76 |
+
proverbs_indexes = [proverb_to_index[proverb] for proverb in test_proverbs]
|
| 77 |
+
expected_proverbs_embeddings = [embeddings[i] for i in proverbs_indexes]
|
| 78 |
+
|
| 79 |
+
# Compute average distance and variance between actual and expected proverbs
|
| 80 |
+
distances = faiss.pairwise_distances(
|
| 81 |
+
actual_proverbs_embeddings, expected_proverbs_embeddings, metric=index.metric_type)
|
| 82 |
+
avg_distance = distances.mean()
|
| 83 |
+
var_distance = distances.var()
|
| 84 |
+
logger.info(
|
| 85 |
+
f"Computed average distance between actual and expected proverbs: {avg_distance}")
|
| 86 |
+
logger.info(
|
| 87 |
+
f"Computed variance of distances between actual and expected proverbs: {var_distance}")
|
| 88 |
+
|
| 89 |
+
test_results = {
|
| 90 |
+
"model": model_name,
|
| 91 |
+
"index_type": index_type.__name__,
|
| 92 |
+
"prompts_test_set_length": len(prompts_test_set),
|
| 93 |
+
"avg_distance": float(avg_distance),
|
| 94 |
+
"var_distance": float(var_distance),
|
| 95 |
+
"map": map[0],
|
| 96 |
+
"map_fields": map[0].split("_"),
|
| 97 |
+
"remarks": remarks,
|
| 98 |
+
"pooling_method": pooling_method,
|
| 99 |
+
}
|
| 100 |
+
|
| 101 |
+
return test_results
|
| 102 |
+
|
| 103 |
+
|
| 104 |
+
def generate_unique_id() -> str:
|
| 105 |
+
"""Build a unique identifier including the current timestamp."""
|
| 106 |
+
timestamp = time.strftime("%Y%m%d_%H%M%S")
|
| 107 |
+
id = timestamp
|
| 108 |
+
return id
|
| 109 |
+
|
| 110 |
+
|
| 111 |
+
if __name__ == "__main__":
|
| 112 |
+
# Hyperparameters to combine and iterate over
|
| 113 |
+
MODELS = models.MODELS
|
| 114 |
+
PROVERB_FIELD_MAPS = {
|
| 115 |
+
"proverb_sentiment_usage": datasets.default_proverb_fields_selection
|
| 116 |
+
}
|
| 117 |
+
INDEX_TYPES = [indexes.DEFAULT_INDEX_TYPE]
|
| 118 |
+
POOLING_METHODS = [models.DEFAULT_POOLING_METHOD]
|
| 119 |
+
|
| 120 |
+
remarks = "ALL hyperparameters combinations, this is going to take a while..."
|
| 121 |
+
|
| 122 |
+
def log_test_case(test_number: int, test_case_id: str) -> str:
|
| 123 |
+
"""Local function to log the test case information using locally defined variables."""
|
| 124 |
+
|
| 125 |
+
# Calculate the maximum lengths for formatting
|
| 126 |
+
max_len_models = max(len(model) for model in MODELS)
|
| 127 |
+
max_len_maps = max(len(map) for map in PROVERB_FIELD_MAPS.keys())
|
| 128 |
+
max_len_index_types = max(len(index_type.__name__)
|
| 129 |
+
for index_type in INDEX_TYPES)
|
| 130 |
+
max_len_pooling_methods = max(len(pooling_method)
|
| 131 |
+
for pooling_method in POOLING_METHODS)
|
| 132 |
+
total_number_tests = len(
|
| 133 |
+
MODELS) * len(PROVERB_FIELD_MAPS) * len(INDEX_TYPES) * len(POOLING_METHODS)
|
| 134 |
+
max_len_test_number = len(str(total_number_tests))
|
| 135 |
+
|
| 136 |
+
# Log the test case information
|
| 137 |
+
logger.info(
|
| 138 |
+
f"({str(test_number).rjust(max_len_test_number)}/{total_number_tests}) " +
|
| 139 |
+
f"Test case {test_case_id}: " +
|
| 140 |
+
f"model = {model_name.ljust(max_len_models)}, " +
|
| 141 |
+
f"index type = {index_type.__name__.ljust(max_len_index_types)}, " +
|
| 142 |
+
f"map = {map[0].ljust(max_len_maps)}, " +
|
| 143 |
+
f"pooling = {pooling_method.ljust(max_len_pooling_methods)} "
|
| 144 |
+
)
|
| 145 |
+
|
| 146 |
+
tokenizer, model, proverbs, prompts_test_set = setup()
|
| 147 |
+
|
| 148 |
+
# Set up the test run
|
| 149 |
+
tests_run_id = generate_unique_id()
|
| 150 |
+
run_folder = os.path.join(f"tests_runs", tests_run_id)
|
| 151 |
+
os.makedirs(run_folder)
|
| 152 |
+
tests_run_file = os.path.join(
|
| 153 |
+
run_folder, f"results_test_run_{tests_run_id}.json")
|
| 154 |
+
|
| 155 |
+
# Copy the test set to the run folder for reproducibility
|
| 156 |
+
shutil.copy2(datasets.PROMPTS_TEST_FILE, run_folder)
|
| 157 |
+
|
| 158 |
+
tests_run_results = {}
|
| 159 |
+
test_number = 1
|
| 160 |
+
|
| 161 |
+
for model_name in MODELS:
|
| 162 |
+
model = models.load_model(model_name)
|
| 163 |
+
tokenizer = models.load_tokenizer(model_name)
|
| 164 |
+
for map in PROVERB_FIELD_MAPS.items():
|
| 165 |
+
for pooling_method in POOLING_METHODS:
|
| 166 |
+
for index_type in INDEX_TYPES:
|
| 167 |
+
# Generate unique identifier for the test case
|
| 168 |
+
test_case_id = generate_unique_id()
|
| 169 |
+
|
| 170 |
+
log_test_case(test_number, test_case_id)
|
| 171 |
+
test_case_results = test_distances(
|
| 172 |
+
tokenizer, model, model_name, proverbs, prompts_test_set, map, index_type, pooling_method, remarks
|
| 173 |
+
)
|
| 174 |
+
|
| 175 |
+
# Store test case results into a JSON
|
| 176 |
+
# (backup intermediate results in case of failure)
|
| 177 |
+
test_case_file = os.path.join(
|
| 178 |
+
run_folder, f"results_test_case_{test_case_id}.json")
|
| 179 |
+
with open(test_case_file, "w") as f:
|
| 180 |
+
json.dump(test_case_results, f, indent=2)
|
| 181 |
+
|
| 182 |
+
tests_run_results[test_case_id] = test_case_results
|
| 183 |
+
test_number += 1
|
| 184 |
+
|
| 185 |
+
# Store test run results into a JSON
|
| 186 |
+
with open(tests_run_file, "w") as f:
|
| 187 |
+
json.dump(tests_run_results, f, indent=2)
|
tests_runs/20250421_183539_maps/results_test_run_20250421_183539.json
ADDED
|
@@ -0,0 +1,906 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"20250421_183541": {
|
| 3 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 4 |
+
"index_type": "IndexFlatL2",
|
| 5 |
+
"prompts_test_set_length": 80,
|
| 6 |
+
"avg_distance": 0.3791416883468628,
|
| 7 |
+
"var_distance": 0.004056941717863083,
|
| 8 |
+
"map": "proverb",
|
| 9 |
+
"map_fields": [
|
| 10 |
+
"proverb"
|
| 11 |
+
],
|
| 12 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 13 |
+
"pooling_method": "mean"
|
| 14 |
+
},
|
| 15 |
+
"20250421_183550": {
|
| 16 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 17 |
+
"index_type": "IndexFlatL2",
|
| 18 |
+
"prompts_test_set_length": 80,
|
| 19 |
+
"avg_distance": 0.6370604038238525,
|
| 20 |
+
"var_distance": 0.011043976992368698,
|
| 21 |
+
"map": "themes",
|
| 22 |
+
"map_fields": [
|
| 23 |
+
"themes"
|
| 24 |
+
],
|
| 25 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 26 |
+
"pooling_method": "mean"
|
| 27 |
+
},
|
| 28 |
+
"20250421_183557": {
|
| 29 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 30 |
+
"index_type": "IndexFlatL2",
|
| 31 |
+
"prompts_test_set_length": 80,
|
| 32 |
+
"avg_distance": 0.49145495891571045,
|
| 33 |
+
"var_distance": 0.003330292645841837,
|
| 34 |
+
"map": "sentiment",
|
| 35 |
+
"map_fields": [
|
| 36 |
+
"sentiment"
|
| 37 |
+
],
|
| 38 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 39 |
+
"pooling_method": "mean"
|
| 40 |
+
},
|
| 41 |
+
"20250421_183602": {
|
| 42 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 43 |
+
"index_type": "IndexFlatL2",
|
| 44 |
+
"prompts_test_set_length": 80,
|
| 45 |
+
"avg_distance": 0.33124756813049316,
|
| 46 |
+
"var_distance": 0.0026342717465013266,
|
| 47 |
+
"map": "explanation",
|
| 48 |
+
"map_fields": [
|
| 49 |
+
"explanation"
|
| 50 |
+
],
|
| 51 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 52 |
+
"pooling_method": "mean"
|
| 53 |
+
},
|
| 54 |
+
"20250421_183612": {
|
| 55 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 56 |
+
"index_type": "IndexFlatL2",
|
| 57 |
+
"prompts_test_set_length": 80,
|
| 58 |
+
"avg_distance": 0.3362385630607605,
|
| 59 |
+
"var_distance": 0.002530086087062955,
|
| 60 |
+
"map": "usage",
|
| 61 |
+
"map_fields": [
|
| 62 |
+
"usage"
|
| 63 |
+
],
|
| 64 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 65 |
+
"pooling_method": "mean"
|
| 66 |
+
},
|
| 67 |
+
"20250421_183622": {
|
| 68 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 69 |
+
"index_type": "IndexFlatL2",
|
| 70 |
+
"prompts_test_set_length": 80,
|
| 71 |
+
"avg_distance": 0.49217942357063293,
|
| 72 |
+
"var_distance": 0.008912492543458939,
|
| 73 |
+
"map": "proverb_themes",
|
| 74 |
+
"map_fields": [
|
| 75 |
+
"proverb",
|
| 76 |
+
"themes"
|
| 77 |
+
],
|
| 78 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 79 |
+
"pooling_method": "mean"
|
| 80 |
+
},
|
| 81 |
+
"20250421_183631": {
|
| 82 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 83 |
+
"index_type": "IndexFlatL2",
|
| 84 |
+
"prompts_test_set_length": 80,
|
| 85 |
+
"avg_distance": 0.4185456335544586,
|
| 86 |
+
"var_distance": 0.007271634414792061,
|
| 87 |
+
"map": "proverb_sentiment",
|
| 88 |
+
"map_fields": [
|
| 89 |
+
"proverb",
|
| 90 |
+
"sentiment"
|
| 91 |
+
],
|
| 92 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 93 |
+
"pooling_method": "mean"
|
| 94 |
+
},
|
| 95 |
+
"20250421_183641": {
|
| 96 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 97 |
+
"index_type": "IndexFlatL2",
|
| 98 |
+
"prompts_test_set_length": 80,
|
| 99 |
+
"avg_distance": 0.35088402032852173,
|
| 100 |
+
"var_distance": 0.003060684772208333,
|
| 101 |
+
"map": "proverb_explanation",
|
| 102 |
+
"map_fields": [
|
| 103 |
+
"proverb",
|
| 104 |
+
"explanation"
|
| 105 |
+
],
|
| 106 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 107 |
+
"pooling_method": "mean"
|
| 108 |
+
},
|
| 109 |
+
"20250421_183658": {
|
| 110 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 111 |
+
"index_type": "IndexFlatL2",
|
| 112 |
+
"prompts_test_set_length": 80,
|
| 113 |
+
"avg_distance": 0.34103038907051086,
|
| 114 |
+
"var_distance": 0.0026473423931747675,
|
| 115 |
+
"map": "proverb_usage",
|
| 116 |
+
"map_fields": [
|
| 117 |
+
"proverb",
|
| 118 |
+
"usage"
|
| 119 |
+
],
|
| 120 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 121 |
+
"pooling_method": "mean"
|
| 122 |
+
},
|
| 123 |
+
"20250421_183711": {
|
| 124 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 125 |
+
"index_type": "IndexFlatL2",
|
| 126 |
+
"prompts_test_set_length": 80,
|
| 127 |
+
"avg_distance": 0.6028177738189697,
|
| 128 |
+
"var_distance": 0.010809319093823433,
|
| 129 |
+
"map": "themes_sentiment",
|
| 130 |
+
"map_fields": [
|
| 131 |
+
"themes",
|
| 132 |
+
"sentiment"
|
| 133 |
+
],
|
| 134 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 135 |
+
"pooling_method": "mean"
|
| 136 |
+
},
|
| 137 |
+
"20250421_183720": {
|
| 138 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 139 |
+
"index_type": "IndexFlatL2",
|
| 140 |
+
"prompts_test_set_length": 80,
|
| 141 |
+
"avg_distance": 0.41258788108825684,
|
| 142 |
+
"var_distance": 0.004030563402920961,
|
| 143 |
+
"map": "themes_explanation",
|
| 144 |
+
"map_fields": [
|
| 145 |
+
"themes",
|
| 146 |
+
"explanation"
|
| 147 |
+
],
|
| 148 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 149 |
+
"pooling_method": "mean"
|
| 150 |
+
},
|
| 151 |
+
"20250421_183733": {
|
| 152 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 153 |
+
"index_type": "IndexFlatL2",
|
| 154 |
+
"prompts_test_set_length": 80,
|
| 155 |
+
"avg_distance": 0.4266483187675476,
|
| 156 |
+
"var_distance": 0.004230715334415436,
|
| 157 |
+
"map": "themes_usage",
|
| 158 |
+
"map_fields": [
|
| 159 |
+
"themes",
|
| 160 |
+
"usage"
|
| 161 |
+
],
|
| 162 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 163 |
+
"pooling_method": "mean"
|
| 164 |
+
},
|
| 165 |
+
"20250421_183745": {
|
| 166 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 167 |
+
"index_type": "IndexFlatL2",
|
| 168 |
+
"prompts_test_set_length": 80,
|
| 169 |
+
"avg_distance": 0.34697455167770386,
|
| 170 |
+
"var_distance": 0.0029269896913319826,
|
| 171 |
+
"map": "sentiment_explanation",
|
| 172 |
+
"map_fields": [
|
| 173 |
+
"sentiment",
|
| 174 |
+
"explanation"
|
| 175 |
+
],
|
| 176 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 177 |
+
"pooling_method": "mean"
|
| 178 |
+
},
|
| 179 |
+
"20250421_183800": {
|
| 180 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 181 |
+
"index_type": "IndexFlatL2",
|
| 182 |
+
"prompts_test_set_length": 80,
|
| 183 |
+
"avg_distance": 0.35297176241874695,
|
| 184 |
+
"var_distance": 0.0024212761782109737,
|
| 185 |
+
"map": "sentiment_usage",
|
| 186 |
+
"map_fields": [
|
| 187 |
+
"sentiment",
|
| 188 |
+
"usage"
|
| 189 |
+
],
|
| 190 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 191 |
+
"pooling_method": "mean"
|
| 192 |
+
},
|
| 193 |
+
"20250421_183810": {
|
| 194 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 195 |
+
"index_type": "IndexFlatL2",
|
| 196 |
+
"prompts_test_set_length": 80,
|
| 197 |
+
"avg_distance": 0.3449043929576874,
|
| 198 |
+
"var_distance": 0.0023571148049086332,
|
| 199 |
+
"map": "explanation_usage",
|
| 200 |
+
"map_fields": [
|
| 201 |
+
"explanation",
|
| 202 |
+
"usage"
|
| 203 |
+
],
|
| 204 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 205 |
+
"pooling_method": "mean"
|
| 206 |
+
},
|
| 207 |
+
"20250421_183825": {
|
| 208 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 209 |
+
"index_type": "IndexFlatL2",
|
| 210 |
+
"prompts_test_set_length": 80,
|
| 211 |
+
"avg_distance": 0.4845268130302429,
|
| 212 |
+
"var_distance": 0.007154541090130806,
|
| 213 |
+
"map": "proverb_themes_sentiment",
|
| 214 |
+
"map_fields": [
|
| 215 |
+
"proverb",
|
| 216 |
+
"themes",
|
| 217 |
+
"sentiment"
|
| 218 |
+
],
|
| 219 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 220 |
+
"pooling_method": "mean"
|
| 221 |
+
},
|
| 222 |
+
"20250421_183834": {
|
| 223 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 224 |
+
"index_type": "IndexFlatL2",
|
| 225 |
+
"prompts_test_set_length": 80,
|
| 226 |
+
"avg_distance": 0.43210312724113464,
|
| 227 |
+
"var_distance": 0.005883161909878254,
|
| 228 |
+
"map": "proverb_themes_explanation",
|
| 229 |
+
"map_fields": [
|
| 230 |
+
"proverb",
|
| 231 |
+
"themes",
|
| 232 |
+
"explanation"
|
| 233 |
+
],
|
| 234 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 235 |
+
"pooling_method": "mean"
|
| 236 |
+
},
|
| 237 |
+
"20250421_183850": {
|
| 238 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 239 |
+
"index_type": "IndexFlatL2",
|
| 240 |
+
"prompts_test_set_length": 80,
|
| 241 |
+
"avg_distance": 0.42529061436653137,
|
| 242 |
+
"var_distance": 0.0050746058113873005,
|
| 243 |
+
"map": "proverb_themes_usage",
|
| 244 |
+
"map_fields": [
|
| 245 |
+
"proverb",
|
| 246 |
+
"themes",
|
| 247 |
+
"usage"
|
| 248 |
+
],
|
| 249 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 250 |
+
"pooling_method": "mean"
|
| 251 |
+
},
|
| 252 |
+
"20250421_183904": {
|
| 253 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 254 |
+
"index_type": "IndexFlatL2",
|
| 255 |
+
"prompts_test_set_length": 80,
|
| 256 |
+
"avg_distance": 0.35188353061676025,
|
| 257 |
+
"var_distance": 0.0034640301018953323,
|
| 258 |
+
"map": "proverb_sentiment_explanation",
|
| 259 |
+
"map_fields": [
|
| 260 |
+
"proverb",
|
| 261 |
+
"sentiment",
|
| 262 |
+
"explanation"
|
| 263 |
+
],
|
| 264 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 265 |
+
"pooling_method": "mean"
|
| 266 |
+
},
|
| 267 |
+
"20250421_183921": {
|
| 268 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 269 |
+
"index_type": "IndexFlatL2",
|
| 270 |
+
"prompts_test_set_length": 80,
|
| 271 |
+
"avg_distance": 0.34581702947616577,
|
| 272 |
+
"var_distance": 0.0030483382288366556,
|
| 273 |
+
"map": "proverb_sentiment_usage",
|
| 274 |
+
"map_fields": [
|
| 275 |
+
"proverb",
|
| 276 |
+
"sentiment",
|
| 277 |
+
"usage"
|
| 278 |
+
],
|
| 279 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 280 |
+
"pooling_method": "mean"
|
| 281 |
+
},
|
| 282 |
+
"20250421_183935": {
|
| 283 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 284 |
+
"index_type": "IndexFlatL2",
|
| 285 |
+
"prompts_test_set_length": 80,
|
| 286 |
+
"avg_distance": 0.3525049686431885,
|
| 287 |
+
"var_distance": 0.0024125524796545506,
|
| 288 |
+
"map": "proverb_explanation_usage",
|
| 289 |
+
"map_fields": [
|
| 290 |
+
"proverb",
|
| 291 |
+
"explanation",
|
| 292 |
+
"usage"
|
| 293 |
+
],
|
| 294 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 295 |
+
"pooling_method": "mean"
|
| 296 |
+
},
|
| 297 |
+
"20250421_183955": {
|
| 298 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 299 |
+
"index_type": "IndexFlatL2",
|
| 300 |
+
"prompts_test_set_length": 80,
|
| 301 |
+
"avg_distance": 0.45778214931488037,
|
| 302 |
+
"var_distance": 0.005367530509829521,
|
| 303 |
+
"map": "themes_sentiment_explanation",
|
| 304 |
+
"map_fields": [
|
| 305 |
+
"themes",
|
| 306 |
+
"sentiment",
|
| 307 |
+
"explanation"
|
| 308 |
+
],
|
| 309 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 310 |
+
"pooling_method": "mean"
|
| 311 |
+
},
|
| 312 |
+
"20250421_184008": {
|
| 313 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 314 |
+
"index_type": "IndexFlatL2",
|
| 315 |
+
"prompts_test_set_length": 80,
|
| 316 |
+
"avg_distance": 0.4575955271720886,
|
| 317 |
+
"var_distance": 0.004186298698186874,
|
| 318 |
+
"map": "themes_sentiment_usage",
|
| 319 |
+
"map_fields": [
|
| 320 |
+
"themes",
|
| 321 |
+
"sentiment",
|
| 322 |
+
"usage"
|
| 323 |
+
],
|
| 324 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 325 |
+
"pooling_method": "mean"
|
| 326 |
+
},
|
| 327 |
+
"20250421_184025": {
|
| 328 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 329 |
+
"index_type": "IndexFlatL2",
|
| 330 |
+
"prompts_test_set_length": 80,
|
| 331 |
+
"avg_distance": 0.40655121207237244,
|
| 332 |
+
"var_distance": 0.0035349917598068714,
|
| 333 |
+
"map": "themes_explanation_usage",
|
| 334 |
+
"map_fields": [
|
| 335 |
+
"themes",
|
| 336 |
+
"explanation",
|
| 337 |
+
"usage"
|
| 338 |
+
],
|
| 339 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 340 |
+
"pooling_method": "mean"
|
| 341 |
+
},
|
| 342 |
+
"20250421_184046": {
|
| 343 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 344 |
+
"index_type": "IndexFlatL2",
|
| 345 |
+
"prompts_test_set_length": 80,
|
| 346 |
+
"avg_distance": 0.3600430190563202,
|
| 347 |
+
"var_distance": 0.0024281726218760014,
|
| 348 |
+
"map": "sentiment_explanation_usage",
|
| 349 |
+
"map_fields": [
|
| 350 |
+
"sentiment",
|
| 351 |
+
"explanation",
|
| 352 |
+
"usage"
|
| 353 |
+
],
|
| 354 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 355 |
+
"pooling_method": "mean"
|
| 356 |
+
},
|
| 357 |
+
"20250421_184104": {
|
| 358 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 359 |
+
"index_type": "IndexFlatL2",
|
| 360 |
+
"prompts_test_set_length": 80,
|
| 361 |
+
"avg_distance": 0.4239869713783264,
|
| 362 |
+
"var_distance": 0.004697322845458984,
|
| 363 |
+
"map": "proverb_themes_sentiment_explanation",
|
| 364 |
+
"map_fields": [
|
| 365 |
+
"proverb",
|
| 366 |
+
"themes",
|
| 367 |
+
"sentiment",
|
| 368 |
+
"explanation"
|
| 369 |
+
],
|
| 370 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 371 |
+
"pooling_method": "mean"
|
| 372 |
+
},
|
| 373 |
+
"20250421_184119": {
|
| 374 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 375 |
+
"index_type": "IndexFlatL2",
|
| 376 |
+
"prompts_test_set_length": 80,
|
| 377 |
+
"avg_distance": 0.41822463274002075,
|
| 378 |
+
"var_distance": 0.003969915676862001,
|
| 379 |
+
"map": "proverb_themes_sentiment_usage",
|
| 380 |
+
"map_fields": [
|
| 381 |
+
"proverb",
|
| 382 |
+
"themes",
|
| 383 |
+
"sentiment",
|
| 384 |
+
"usage"
|
| 385 |
+
],
|
| 386 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 387 |
+
"pooling_method": "mean"
|
| 388 |
+
},
|
| 389 |
+
"20250421_184134": {
|
| 390 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 391 |
+
"index_type": "IndexFlatL2",
|
| 392 |
+
"prompts_test_set_length": 80,
|
| 393 |
+
"avg_distance": 0.4041880667209625,
|
| 394 |
+
"var_distance": 0.0037264481652528048,
|
| 395 |
+
"map": "proverb_themes_explanation_usage",
|
| 396 |
+
"map_fields": [
|
| 397 |
+
"proverb",
|
| 398 |
+
"themes",
|
| 399 |
+
"explanation",
|
| 400 |
+
"usage"
|
| 401 |
+
],
|
| 402 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 403 |
+
"pooling_method": "mean"
|
| 404 |
+
},
|
| 405 |
+
"20250421_184156": {
|
| 406 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 407 |
+
"index_type": "IndexFlatL2",
|
| 408 |
+
"prompts_test_set_length": 80,
|
| 409 |
+
"avg_distance": 0.3571434020996094,
|
| 410 |
+
"var_distance": 0.0025321487337350845,
|
| 411 |
+
"map": "proverb_sentiment_explanation_usage",
|
| 412 |
+
"map_fields": [
|
| 413 |
+
"proverb",
|
| 414 |
+
"sentiment",
|
| 415 |
+
"explanation",
|
| 416 |
+
"usage"
|
| 417 |
+
],
|
| 418 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 419 |
+
"pooling_method": "mean"
|
| 420 |
+
},
|
| 421 |
+
"20250421_184217": {
|
| 422 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 423 |
+
"index_type": "IndexFlatL2",
|
| 424 |
+
"prompts_test_set_length": 80,
|
| 425 |
+
"avg_distance": 0.4234916567802429,
|
| 426 |
+
"var_distance": 0.003560721641406417,
|
| 427 |
+
"map": "themes_sentiment_explanation_usage",
|
| 428 |
+
"map_fields": [
|
| 429 |
+
"themes",
|
| 430 |
+
"sentiment",
|
| 431 |
+
"explanation",
|
| 432 |
+
"usage"
|
| 433 |
+
],
|
| 434 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 435 |
+
"pooling_method": "mean"
|
| 436 |
+
},
|
| 437 |
+
"20250421_184236": {
|
| 438 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 439 |
+
"index_type": "IndexFlatL2",
|
| 440 |
+
"prompts_test_set_length": 80,
|
| 441 |
+
"avg_distance": 0.4021625518798828,
|
| 442 |
+
"var_distance": 0.0031073391437530518,
|
| 443 |
+
"map": "proverb_themes_sentiment_explanation_usage",
|
| 444 |
+
"map_fields": [
|
| 445 |
+
"proverb",
|
| 446 |
+
"themes",
|
| 447 |
+
"sentiment",
|
| 448 |
+
"explanation",
|
| 449 |
+
"usage"
|
| 450 |
+
],
|
| 451 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 452 |
+
"pooling_method": "mean"
|
| 453 |
+
},
|
| 454 |
+
"20250421_184307": {
|
| 455 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 456 |
+
"index_type": "IndexFlatL2",
|
| 457 |
+
"prompts_test_set_length": 80,
|
| 458 |
+
"avg_distance": 0.08793239295482635,
|
| 459 |
+
"var_distance": 0.011847684159874916,
|
| 460 |
+
"map": "proverb",
|
| 461 |
+
"map_fields": [
|
| 462 |
+
"proverb"
|
| 463 |
+
],
|
| 464 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 465 |
+
"pooling_method": "mean"
|
| 466 |
+
},
|
| 467 |
+
"20250421_184316": {
|
| 468 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 469 |
+
"index_type": "IndexFlatL2",
|
| 470 |
+
"prompts_test_set_length": 80,
|
| 471 |
+
"avg_distance": 0.07392635196447372,
|
| 472 |
+
"var_distance": 0.0077307759784162045,
|
| 473 |
+
"map": "themes",
|
| 474 |
+
"map_fields": [
|
| 475 |
+
"themes"
|
| 476 |
+
],
|
| 477 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 478 |
+
"pooling_method": "mean"
|
| 479 |
+
},
|
| 480 |
+
"20250421_184323": {
|
| 481 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 482 |
+
"index_type": "IndexFlatL2",
|
| 483 |
+
"prompts_test_set_length": 80,
|
| 484 |
+
"avg_distance": 0.1869734525680542,
|
| 485 |
+
"var_distance": 0.03088298812508583,
|
| 486 |
+
"map": "sentiment",
|
| 487 |
+
"map_fields": [
|
| 488 |
+
"sentiment"
|
| 489 |
+
],
|
| 490 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 491 |
+
"pooling_method": "mean"
|
| 492 |
+
},
|
| 493 |
+
"20250421_184328": {
|
| 494 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 495 |
+
"index_type": "IndexFlatL2",
|
| 496 |
+
"prompts_test_set_length": 80,
|
| 497 |
+
"avg_distance": 0.06268821656703949,
|
| 498 |
+
"var_distance": 0.006070807110518217,
|
| 499 |
+
"map": "explanation",
|
| 500 |
+
"map_fields": [
|
| 501 |
+
"explanation"
|
| 502 |
+
],
|
| 503 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 504 |
+
"pooling_method": "mean"
|
| 505 |
+
},
|
| 506 |
+
"20250421_184340": {
|
| 507 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 508 |
+
"index_type": "IndexFlatL2",
|
| 509 |
+
"prompts_test_set_length": 80,
|
| 510 |
+
"avg_distance": 0.1029096245765686,
|
| 511 |
+
"var_distance": 0.013397054746747017,
|
| 512 |
+
"map": "usage",
|
| 513 |
+
"map_fields": [
|
| 514 |
+
"usage"
|
| 515 |
+
],
|
| 516 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 517 |
+
"pooling_method": "mean"
|
| 518 |
+
},
|
| 519 |
+
"20250421_184350": {
|
| 520 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 521 |
+
"index_type": "IndexFlatL2",
|
| 522 |
+
"prompts_test_set_length": 80,
|
| 523 |
+
"avg_distance": 0.061103705316782,
|
| 524 |
+
"var_distance": 0.0054617952555418015,
|
| 525 |
+
"map": "proverb_themes",
|
| 526 |
+
"map_fields": [
|
| 527 |
+
"proverb",
|
| 528 |
+
"themes"
|
| 529 |
+
],
|
| 530 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 531 |
+
"pooling_method": "mean"
|
| 532 |
+
},
|
| 533 |
+
"20250421_184400": {
|
| 534 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 535 |
+
"index_type": "IndexFlatL2",
|
| 536 |
+
"prompts_test_set_length": 80,
|
| 537 |
+
"avg_distance": 0.0819675624370575,
|
| 538 |
+
"var_distance": 0.010567250661551952,
|
| 539 |
+
"map": "proverb_sentiment",
|
| 540 |
+
"map_fields": [
|
| 541 |
+
"proverb",
|
| 542 |
+
"sentiment"
|
| 543 |
+
],
|
| 544 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 545 |
+
"pooling_method": "mean"
|
| 546 |
+
},
|
| 547 |
+
"20250421_184409": {
|
| 548 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 549 |
+
"index_type": "IndexFlatL2",
|
| 550 |
+
"prompts_test_set_length": 80,
|
| 551 |
+
"avg_distance": 0.08502128720283508,
|
| 552 |
+
"var_distance": 0.010671218857169151,
|
| 553 |
+
"map": "proverb_explanation",
|
| 554 |
+
"map_fields": [
|
| 555 |
+
"proverb",
|
| 556 |
+
"explanation"
|
| 557 |
+
],
|
| 558 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 559 |
+
"pooling_method": "mean"
|
| 560 |
+
},
|
| 561 |
+
"20250421_184423": {
|
| 562 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 563 |
+
"index_type": "IndexFlatL2",
|
| 564 |
+
"prompts_test_set_length": 80,
|
| 565 |
+
"avg_distance": 0.06733336299657822,
|
| 566 |
+
"var_distance": 0.0067588090896606445,
|
| 567 |
+
"map": "proverb_usage",
|
| 568 |
+
"map_fields": [
|
| 569 |
+
"proverb",
|
| 570 |
+
"usage"
|
| 571 |
+
],
|
| 572 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 573 |
+
"pooling_method": "mean"
|
| 574 |
+
},
|
| 575 |
+
"20250421_184436": {
|
| 576 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 577 |
+
"index_type": "IndexFlatL2",
|
| 578 |
+
"prompts_test_set_length": 80,
|
| 579 |
+
"avg_distance": 0.07368369400501251,
|
| 580 |
+
"var_distance": 0.010688500478863716,
|
| 581 |
+
"map": "themes_sentiment",
|
| 582 |
+
"map_fields": [
|
| 583 |
+
"themes",
|
| 584 |
+
"sentiment"
|
| 585 |
+
],
|
| 586 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 587 |
+
"pooling_method": "mean"
|
| 588 |
+
},
|
| 589 |
+
"20250421_184443": {
|
| 590 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 591 |
+
"index_type": "IndexFlatL2",
|
| 592 |
+
"prompts_test_set_length": 80,
|
| 593 |
+
"avg_distance": 0.08933711796998978,
|
| 594 |
+
"var_distance": 0.014037378132343292,
|
| 595 |
+
"map": "themes_explanation",
|
| 596 |
+
"map_fields": [
|
| 597 |
+
"themes",
|
| 598 |
+
"explanation"
|
| 599 |
+
],
|
| 600 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 601 |
+
"pooling_method": "mean"
|
| 602 |
+
},
|
| 603 |
+
"20250421_184457": {
|
| 604 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 605 |
+
"index_type": "IndexFlatL2",
|
| 606 |
+
"prompts_test_set_length": 80,
|
| 607 |
+
"avg_distance": 0.12320074439048767,
|
| 608 |
+
"var_distance": 0.02022252231836319,
|
| 609 |
+
"map": "themes_usage",
|
| 610 |
+
"map_fields": [
|
| 611 |
+
"themes",
|
| 612 |
+
"usage"
|
| 613 |
+
],
|
| 614 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 615 |
+
"pooling_method": "mean"
|
| 616 |
+
},
|
| 617 |
+
"20250421_184509": {
|
| 618 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 619 |
+
"index_type": "IndexFlatL2",
|
| 620 |
+
"prompts_test_set_length": 80,
|
| 621 |
+
"avg_distance": 0.06253966689109802,
|
| 622 |
+
"var_distance": 0.005820302292704582,
|
| 623 |
+
"map": "sentiment_explanation",
|
| 624 |
+
"map_fields": [
|
| 625 |
+
"sentiment",
|
| 626 |
+
"explanation"
|
| 627 |
+
],
|
| 628 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 629 |
+
"pooling_method": "mean"
|
| 630 |
+
},
|
| 631 |
+
"20250421_184521": {
|
| 632 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 633 |
+
"index_type": "IndexFlatL2",
|
| 634 |
+
"prompts_test_set_length": 80,
|
| 635 |
+
"avg_distance": 0.10977248102426529,
|
| 636 |
+
"var_distance": 0.011955869384109974,
|
| 637 |
+
"map": "sentiment_usage",
|
| 638 |
+
"map_fields": [
|
| 639 |
+
"sentiment",
|
| 640 |
+
"usage"
|
| 641 |
+
],
|
| 642 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 643 |
+
"pooling_method": "mean"
|
| 644 |
+
},
|
| 645 |
+
"20250421_184532": {
|
| 646 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 647 |
+
"index_type": "IndexFlatL2",
|
| 648 |
+
"prompts_test_set_length": 80,
|
| 649 |
+
"avg_distance": 0.15491245687007904,
|
| 650 |
+
"var_distance": 0.027762046083807945,
|
| 651 |
+
"map": "explanation_usage",
|
| 652 |
+
"map_fields": [
|
| 653 |
+
"explanation",
|
| 654 |
+
"usage"
|
| 655 |
+
],
|
| 656 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 657 |
+
"pooling_method": "mean"
|
| 658 |
+
},
|
| 659 |
+
"20250421_184549": {
|
| 660 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 661 |
+
"index_type": "IndexFlatL2",
|
| 662 |
+
"prompts_test_set_length": 80,
|
| 663 |
+
"avg_distance": 0.06401202827692032,
|
| 664 |
+
"var_distance": 0.006882899906486273,
|
| 665 |
+
"map": "proverb_themes_sentiment",
|
| 666 |
+
"map_fields": [
|
| 667 |
+
"proverb",
|
| 668 |
+
"themes",
|
| 669 |
+
"sentiment"
|
| 670 |
+
],
|
| 671 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 672 |
+
"pooling_method": "mean"
|
| 673 |
+
},
|
| 674 |
+
"20250421_184559": {
|
| 675 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 676 |
+
"index_type": "IndexFlatL2",
|
| 677 |
+
"prompts_test_set_length": 80,
|
| 678 |
+
"avg_distance": 0.0867539495229721,
|
| 679 |
+
"var_distance": 0.011152057908475399,
|
| 680 |
+
"map": "proverb_themes_explanation",
|
| 681 |
+
"map_fields": [
|
| 682 |
+
"proverb",
|
| 683 |
+
"themes",
|
| 684 |
+
"explanation"
|
| 685 |
+
],
|
| 686 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 687 |
+
"pooling_method": "mean"
|
| 688 |
+
},
|
| 689 |
+
"20250421_184615": {
|
| 690 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 691 |
+
"index_type": "IndexFlatL2",
|
| 692 |
+
"prompts_test_set_length": 80,
|
| 693 |
+
"avg_distance": 0.08302739262580872,
|
| 694 |
+
"var_distance": 0.010598292574286461,
|
| 695 |
+
"map": "proverb_themes_usage",
|
| 696 |
+
"map_fields": [
|
| 697 |
+
"proverb",
|
| 698 |
+
"themes",
|
| 699 |
+
"usage"
|
| 700 |
+
],
|
| 701 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 702 |
+
"pooling_method": "mean"
|
| 703 |
+
},
|
| 704 |
+
"20250421_184629": {
|
| 705 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 706 |
+
"index_type": "IndexFlatL2",
|
| 707 |
+
"prompts_test_set_length": 80,
|
| 708 |
+
"avg_distance": 0.08669276535511017,
|
| 709 |
+
"var_distance": 0.011783267371356487,
|
| 710 |
+
"map": "proverb_sentiment_explanation",
|
| 711 |
+
"map_fields": [
|
| 712 |
+
"proverb",
|
| 713 |
+
"sentiment",
|
| 714 |
+
"explanation"
|
| 715 |
+
],
|
| 716 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 717 |
+
"pooling_method": "mean"
|
| 718 |
+
},
|
| 719 |
+
"20250421_184644": {
|
| 720 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 721 |
+
"index_type": "IndexFlatL2",
|
| 722 |
+
"prompts_test_set_length": 80,
|
| 723 |
+
"avg_distance": 0.07192985713481903,
|
| 724 |
+
"var_distance": 0.007489962503314018,
|
| 725 |
+
"map": "proverb_sentiment_usage",
|
| 726 |
+
"map_fields": [
|
| 727 |
+
"proverb",
|
| 728 |
+
"sentiment",
|
| 729 |
+
"usage"
|
| 730 |
+
],
|
| 731 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 732 |
+
"pooling_method": "mean"
|
| 733 |
+
},
|
| 734 |
+
"20250421_184658": {
|
| 735 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 736 |
+
"index_type": "IndexFlatL2",
|
| 737 |
+
"prompts_test_set_length": 80,
|
| 738 |
+
"avg_distance": 0.13168814778327942,
|
| 739 |
+
"var_distance": 0.020178651437163353,
|
| 740 |
+
"map": "proverb_explanation_usage",
|
| 741 |
+
"map_fields": [
|
| 742 |
+
"proverb",
|
| 743 |
+
"explanation",
|
| 744 |
+
"usage"
|
| 745 |
+
],
|
| 746 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 747 |
+
"pooling_method": "mean"
|
| 748 |
+
},
|
| 749 |
+
"20250421_184717": {
|
| 750 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 751 |
+
"index_type": "IndexFlatL2",
|
| 752 |
+
"prompts_test_set_length": 80,
|
| 753 |
+
"avg_distance": 0.09910964965820312,
|
| 754 |
+
"var_distance": 0.014508042484521866,
|
| 755 |
+
"map": "themes_sentiment_explanation",
|
| 756 |
+
"map_fields": [
|
| 757 |
+
"themes",
|
| 758 |
+
"sentiment",
|
| 759 |
+
"explanation"
|
| 760 |
+
],
|
| 761 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 762 |
+
"pooling_method": "mean"
|
| 763 |
+
},
|
| 764 |
+
"20250421_184734": {
|
| 765 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 766 |
+
"index_type": "IndexFlatL2",
|
| 767 |
+
"prompts_test_set_length": 80,
|
| 768 |
+
"avg_distance": 0.14658302068710327,
|
| 769 |
+
"var_distance": 0.02491208352148533,
|
| 770 |
+
"map": "themes_sentiment_usage",
|
| 771 |
+
"map_fields": [
|
| 772 |
+
"themes",
|
| 773 |
+
"sentiment",
|
| 774 |
+
"usage"
|
| 775 |
+
],
|
| 776 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 777 |
+
"pooling_method": "mean"
|
| 778 |
+
},
|
| 779 |
+
"20250421_184746": {
|
| 780 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 781 |
+
"index_type": "IndexFlatL2",
|
| 782 |
+
"prompts_test_set_length": 80,
|
| 783 |
+
"avg_distance": 0.17632853984832764,
|
| 784 |
+
"var_distance": 0.029088711366057396,
|
| 785 |
+
"map": "themes_explanation_usage",
|
| 786 |
+
"map_fields": [
|
| 787 |
+
"themes",
|
| 788 |
+
"explanation",
|
| 789 |
+
"usage"
|
| 790 |
+
],
|
| 791 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 792 |
+
"pooling_method": "mean"
|
| 793 |
+
},
|
| 794 |
+
"20250421_184803": {
|
| 795 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 796 |
+
"index_type": "IndexFlatL2",
|
| 797 |
+
"prompts_test_set_length": 80,
|
| 798 |
+
"avg_distance": 0.1599045991897583,
|
| 799 |
+
"var_distance": 0.02734311670064926,
|
| 800 |
+
"map": "sentiment_explanation_usage",
|
| 801 |
+
"map_fields": [
|
| 802 |
+
"sentiment",
|
| 803 |
+
"explanation",
|
| 804 |
+
"usage"
|
| 805 |
+
],
|
| 806 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 807 |
+
"pooling_method": "mean"
|
| 808 |
+
},
|
| 809 |
+
"20250421_184823": {
|
| 810 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 811 |
+
"index_type": "IndexFlatL2",
|
| 812 |
+
"prompts_test_set_length": 80,
|
| 813 |
+
"avg_distance": 0.09978899359703064,
|
| 814 |
+
"var_distance": 0.013416523113846779,
|
| 815 |
+
"map": "proverb_themes_sentiment_explanation",
|
| 816 |
+
"map_fields": [
|
| 817 |
+
"proverb",
|
| 818 |
+
"themes",
|
| 819 |
+
"sentiment",
|
| 820 |
+
"explanation"
|
| 821 |
+
],
|
| 822 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 823 |
+
"pooling_method": "mean"
|
| 824 |
+
},
|
| 825 |
+
"20250421_184840": {
|
| 826 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 827 |
+
"index_type": "IndexFlatL2",
|
| 828 |
+
"prompts_test_set_length": 80,
|
| 829 |
+
"avg_distance": 0.09943566471338272,
|
| 830 |
+
"var_distance": 0.012451882474124432,
|
| 831 |
+
"map": "proverb_themes_sentiment_usage",
|
| 832 |
+
"map_fields": [
|
| 833 |
+
"proverb",
|
| 834 |
+
"themes",
|
| 835 |
+
"sentiment",
|
| 836 |
+
"usage"
|
| 837 |
+
],
|
| 838 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 839 |
+
"pooling_method": "mean"
|
| 840 |
+
},
|
| 841 |
+
"20250421_184855": {
|
| 842 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 843 |
+
"index_type": "IndexFlatL2",
|
| 844 |
+
"prompts_test_set_length": 80,
|
| 845 |
+
"avg_distance": 0.14704512059688568,
|
| 846 |
+
"var_distance": 0.02114211581647396,
|
| 847 |
+
"map": "proverb_themes_explanation_usage",
|
| 848 |
+
"map_fields": [
|
| 849 |
+
"proverb",
|
| 850 |
+
"themes",
|
| 851 |
+
"explanation",
|
| 852 |
+
"usage"
|
| 853 |
+
],
|
| 854 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 855 |
+
"pooling_method": "mean"
|
| 856 |
+
},
|
| 857 |
+
"20250421_184916": {
|
| 858 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 859 |
+
"index_type": "IndexFlatL2",
|
| 860 |
+
"prompts_test_set_length": 80,
|
| 861 |
+
"avg_distance": 0.13551250100135803,
|
| 862 |
+
"var_distance": 0.019405102357268333,
|
| 863 |
+
"map": "proverb_sentiment_explanation_usage",
|
| 864 |
+
"map_fields": [
|
| 865 |
+
"proverb",
|
| 866 |
+
"sentiment",
|
| 867 |
+
"explanation",
|
| 868 |
+
"usage"
|
| 869 |
+
],
|
| 870 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 871 |
+
"pooling_method": "mean"
|
| 872 |
+
},
|
| 873 |
+
"20250421_184940": {
|
| 874 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 875 |
+
"index_type": "IndexFlatL2",
|
| 876 |
+
"prompts_test_set_length": 80,
|
| 877 |
+
"avg_distance": 0.1790611296892166,
|
| 878 |
+
"var_distance": 0.027481362223625183,
|
| 879 |
+
"map": "themes_sentiment_explanation_usage",
|
| 880 |
+
"map_fields": [
|
| 881 |
+
"themes",
|
| 882 |
+
"sentiment",
|
| 883 |
+
"explanation",
|
| 884 |
+
"usage"
|
| 885 |
+
],
|
| 886 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 887 |
+
"pooling_method": "mean"
|
| 888 |
+
},
|
| 889 |
+
"20250421_184959": {
|
| 890 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 891 |
+
"index_type": "IndexFlatL2",
|
| 892 |
+
"prompts_test_set_length": 80,
|
| 893 |
+
"avg_distance": 0.1610860973596573,
|
| 894 |
+
"var_distance": 0.022566508501768112,
|
| 895 |
+
"map": "proverb_themes_sentiment_explanation_usage",
|
| 896 |
+
"map_fields": [
|
| 897 |
+
"proverb",
|
| 898 |
+
"themes",
|
| 899 |
+
"sentiment",
|
| 900 |
+
"explanation",
|
| 901 |
+
"usage"
|
| 902 |
+
],
|
| 903 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed index type and pooling method) for field map selection.",
|
| 904 |
+
"pooling_method": "mean"
|
| 905 |
+
}
|
| 906 |
+
}
|
tests_runs/20250421_185513_tuning/results_test_run_20250421_185513.json
ADDED
|
@@ -0,0 +1,182 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"20250421_185515": {
|
| 3 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 4 |
+
"index_type": "IndexFlatL2",
|
| 5 |
+
"prompts_test_set_length": 80,
|
| 6 |
+
"avg_distance": 0.49217942357063293,
|
| 7 |
+
"var_distance": 0.008912492543458939,
|
| 8 |
+
"map": "proverb_sentiment_usage",
|
| 9 |
+
"map_fields": [
|
| 10 |
+
"proverb",
|
| 11 |
+
"sentiment",
|
| 12 |
+
"usage"
|
| 13 |
+
],
|
| 14 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 15 |
+
"pooling_method": "mean"
|
| 16 |
+
},
|
| 17 |
+
"20250421_185526": {
|
| 18 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 19 |
+
"index_type": "IndexFlatIP",
|
| 20 |
+
"prompts_test_set_length": 80,
|
| 21 |
+
"avg_distance": 0.7539103031158447,
|
| 22 |
+
"var_distance": 0.0022281226702034473,
|
| 23 |
+
"map": "proverb_sentiment_usage",
|
| 24 |
+
"map_fields": [
|
| 25 |
+
"proverb",
|
| 26 |
+
"sentiment",
|
| 27 |
+
"usage"
|
| 28 |
+
],
|
| 29 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 30 |
+
"pooling_method": "mean"
|
| 31 |
+
},
|
| 32 |
+
"20250421_185537": {
|
| 33 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 34 |
+
"index_type": "IndexFlatL2",
|
| 35 |
+
"prompts_test_set_length": 80,
|
| 36 |
+
"avg_distance": 0.2621341049671173,
|
| 37 |
+
"var_distance": 0.0012911176308989525,
|
| 38 |
+
"map": "proverb_sentiment_usage",
|
| 39 |
+
"map_fields": [
|
| 40 |
+
"proverb",
|
| 41 |
+
"sentiment",
|
| 42 |
+
"usage"
|
| 43 |
+
],
|
| 44 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 45 |
+
"pooling_method": "max"
|
| 46 |
+
},
|
| 47 |
+
"20250421_185547": {
|
| 48 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 49 |
+
"index_type": "IndexFlatIP",
|
| 50 |
+
"prompts_test_set_length": 80,
|
| 51 |
+
"avg_distance": 0.8689329624176025,
|
| 52 |
+
"var_distance": 0.0003227795532438904,
|
| 53 |
+
"map": "proverb_sentiment_usage",
|
| 54 |
+
"map_fields": [
|
| 55 |
+
"proverb",
|
| 56 |
+
"sentiment",
|
| 57 |
+
"usage"
|
| 58 |
+
],
|
| 59 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 60 |
+
"pooling_method": "max"
|
| 61 |
+
},
|
| 62 |
+
"20250421_185559": {
|
| 63 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 64 |
+
"index_type": "IndexFlatL2",
|
| 65 |
+
"prompts_test_set_length": 80,
|
| 66 |
+
"avg_distance": 0.05769731104373932,
|
| 67 |
+
"var_distance": 7.417293090838939e-05,
|
| 68 |
+
"map": "proverb_sentiment_usage",
|
| 69 |
+
"map_fields": [
|
| 70 |
+
"proverb",
|
| 71 |
+
"sentiment",
|
| 72 |
+
"usage"
|
| 73 |
+
],
|
| 74 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 75 |
+
"pooling_method": "cls"
|
| 76 |
+
},
|
| 77 |
+
"20250421_185610": {
|
| 78 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 79 |
+
"index_type": "IndexFlatIP",
|
| 80 |
+
"prompts_test_set_length": 80,
|
| 81 |
+
"avg_distance": 0.9711512923240662,
|
| 82 |
+
"var_distance": 1.8543241822044365e-05,
|
| 83 |
+
"map": "proverb_sentiment_usage",
|
| 84 |
+
"map_fields": [
|
| 85 |
+
"proverb",
|
| 86 |
+
"sentiment",
|
| 87 |
+
"usage"
|
| 88 |
+
],
|
| 89 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 90 |
+
"pooling_method": "cls"
|
| 91 |
+
},
|
| 92 |
+
"20250421_185623": {
|
| 93 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 94 |
+
"index_type": "IndexFlatL2",
|
| 95 |
+
"prompts_test_set_length": 80,
|
| 96 |
+
"avg_distance": 0.061103705316782,
|
| 97 |
+
"var_distance": 0.0054617952555418015,
|
| 98 |
+
"map": "proverb_sentiment_usage",
|
| 99 |
+
"map_fields": [
|
| 100 |
+
"proverb",
|
| 101 |
+
"sentiment",
|
| 102 |
+
"usage"
|
| 103 |
+
],
|
| 104 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 105 |
+
"pooling_method": "mean"
|
| 106 |
+
},
|
| 107 |
+
"20250421_185633": {
|
| 108 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 109 |
+
"index_type": "IndexFlatIP",
|
| 110 |
+
"prompts_test_set_length": 80,
|
| 111 |
+
"avg_distance": 0.9694480895996094,
|
| 112 |
+
"var_distance": 0.001365448348224163,
|
| 113 |
+
"map": "proverb_sentiment_usage",
|
| 114 |
+
"map_fields": [
|
| 115 |
+
"proverb",
|
| 116 |
+
"sentiment",
|
| 117 |
+
"usage"
|
| 118 |
+
],
|
| 119 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 120 |
+
"pooling_method": "mean"
|
| 121 |
+
},
|
| 122 |
+
"20250421_185644": {
|
| 123 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 124 |
+
"index_type": "IndexFlatL2",
|
| 125 |
+
"prompts_test_set_length": 80,
|
| 126 |
+
"avg_distance": 0.11899012327194214,
|
| 127 |
+
"var_distance": 0.040872879326343536,
|
| 128 |
+
"map": "proverb_sentiment_usage",
|
| 129 |
+
"map_fields": [
|
| 130 |
+
"proverb",
|
| 131 |
+
"sentiment",
|
| 132 |
+
"usage"
|
| 133 |
+
],
|
| 134 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 135 |
+
"pooling_method": "max"
|
| 136 |
+
},
|
| 137 |
+
"20250421_185653": {
|
| 138 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 139 |
+
"index_type": "IndexFlatIP",
|
| 140 |
+
"prompts_test_set_length": 80,
|
| 141 |
+
"avg_distance": 0.9405049085617065,
|
| 142 |
+
"var_distance": 0.01021821703761816,
|
| 143 |
+
"map": "proverb_sentiment_usage",
|
| 144 |
+
"map_fields": [
|
| 145 |
+
"proverb",
|
| 146 |
+
"sentiment",
|
| 147 |
+
"usage"
|
| 148 |
+
],
|
| 149 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 150 |
+
"pooling_method": "max"
|
| 151 |
+
},
|
| 152 |
+
"20250421_185704": {
|
| 153 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 154 |
+
"index_type": "IndexFlatL2",
|
| 155 |
+
"prompts_test_set_length": 80,
|
| 156 |
+
"avg_distance": 0.06070325896143913,
|
| 157 |
+
"var_distance": 0.005394950974732637,
|
| 158 |
+
"map": "proverb_sentiment_usage",
|
| 159 |
+
"map_fields": [
|
| 160 |
+
"proverb",
|
| 161 |
+
"sentiment",
|
| 162 |
+
"usage"
|
| 163 |
+
],
|
| 164 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 165 |
+
"pooling_method": "cls"
|
| 166 |
+
},
|
| 167 |
+
"20250421_185715": {
|
| 168 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 169 |
+
"index_type": "IndexFlatIP",
|
| 170 |
+
"prompts_test_set_length": 80,
|
| 171 |
+
"avg_distance": 0.9696483612060547,
|
| 172 |
+
"var_distance": 0.001348737976513803,
|
| 173 |
+
"map": "proverb_sentiment_usage",
|
| 174 |
+
"map_fields": [
|
| 175 |
+
"proverb",
|
| 176 |
+
"sentiment",
|
| 177 |
+
"usage"
|
| 178 |
+
],
|
| 179 |
+
"remarks": "Only tested on a subset of hyperparameters (2 models, fixed field map) for secondary hyperparameter selection.",
|
| 180 |
+
"pooling_method": "cls"
|
| 181 |
+
}
|
| 182 |
+
}
|
tests_runs/20250421_190552_model/results_test_run_20250421_190552.json
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"20250421_190553": {
|
| 3 |
+
"model": "projecte-aina/roberta-base-ca-v2",
|
| 4 |
+
"index_type": "IndexFlatL2",
|
| 5 |
+
"prompts_test_set_length": 80,
|
| 6 |
+
"avg_distance": 0.05769731104373932,
|
| 7 |
+
"var_distance": 7.417293090838939e-05,
|
| 8 |
+
"map": "proverb_sentiment_usage",
|
| 9 |
+
"map_fields": [
|
| 10 |
+
"proverb",
|
| 11 |
+
"sentiment",
|
| 12 |
+
"usage"
|
| 13 |
+
],
|
| 14 |
+
"remarks": "Tested on all models with other hyperparameters fixed.",
|
| 15 |
+
"pooling_method": "cls"
|
| 16 |
+
},
|
| 17 |
+
"20250421_190603": {
|
| 18 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-sts",
|
| 19 |
+
"index_type": "IndexFlatL2",
|
| 20 |
+
"prompts_test_set_length": 80,
|
| 21 |
+
"avg_distance": 0.06070325896143913,
|
| 22 |
+
"var_distance": 0.005394950974732637,
|
| 23 |
+
"map": "proverb_sentiment_usage",
|
| 24 |
+
"map_fields": [
|
| 25 |
+
"proverb",
|
| 26 |
+
"sentiment",
|
| 27 |
+
"usage"
|
| 28 |
+
],
|
| 29 |
+
"remarks": "Tested on all models with other hyperparameters fixed.",
|
| 30 |
+
"pooling_method": "cls"
|
| 31 |
+
},
|
| 32 |
+
"20250421_190614": {
|
| 33 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-te",
|
| 34 |
+
"index_type": "IndexFlatL2",
|
| 35 |
+
"prompts_test_set_length": 80,
|
| 36 |
+
"avg_distance": 1.2598824501037598,
|
| 37 |
+
"var_distance": 0.5332778692245483,
|
| 38 |
+
"map": "proverb_sentiment_usage",
|
| 39 |
+
"map_fields": [
|
| 40 |
+
"proverb",
|
| 41 |
+
"sentiment",
|
| 42 |
+
"usage"
|
| 43 |
+
],
|
| 44 |
+
"remarks": "Tested on all models with other hyperparameters fixed.",
|
| 45 |
+
"pooling_method": "cls"
|
| 46 |
+
},
|
| 47 |
+
"20250421_190626": {
|
| 48 |
+
"model": "projecte-aina/roberta-base-ca-v2-cased-tc",
|
| 49 |
+
"index_type": "IndexFlatL2",
|
| 50 |
+
"prompts_test_set_length": 80,
|
| 51 |
+
"avg_distance": 1.2142741680145264,
|
| 52 |
+
"var_distance": 0.2768329381942749,
|
| 53 |
+
"map": "proverb_sentiment_usage",
|
| 54 |
+
"map_fields": [
|
| 55 |
+
"proverb",
|
| 56 |
+
"sentiment",
|
| 57 |
+
"usage"
|
| 58 |
+
],
|
| 59 |
+
"remarks": "Tested on all models with other hyperparameters fixed.",
|
| 60 |
+
"pooling_method": "cls"
|
| 61 |
+
},
|
| 62 |
+
"20250421_190639": {
|
| 63 |
+
"model": "projecte-aina/roberta-large-ca-v2",
|
| 64 |
+
"index_type": "IndexFlatL2",
|
| 65 |
+
"prompts_test_set_length": 80,
|
| 66 |
+
"avg_distance": 0.4151465892791748,
|
| 67 |
+
"var_distance": 0.06712639331817627,
|
| 68 |
+
"map": "proverb_sentiment_usage",
|
| 69 |
+
"map_fields": [
|
| 70 |
+
"proverb",
|
| 71 |
+
"sentiment",
|
| 72 |
+
"usage"
|
| 73 |
+
],
|
| 74 |
+
"remarks": "Tested on all models with other hyperparameters fixed.",
|
| 75 |
+
"pooling_method": "cls"
|
| 76 |
+
}
|
| 77 |
+
}
|