SAVRN
Search Contact SAVRN

SAVRN Model Hub · Comparisons

distilbert-base-uncased-finetuned-sst-2-english vs distilroberta-finetuned-financial-news-sentiment-analysis

Distilbert-base-uncased-finetuned-sst-2-english has 67M parameters and distilroberta-finetuned-financial-news-sentiment-analysis has 82M parameters; both are released under Apache License 2.0; at 16-bit, distilbert-base-uncased-finetuned-sst-2-english needs about 0.2 GB (1x MI300X from $1.85 an hour) and distilroberta-finetuned-financial-news-sentiment-analysis about 0.2 GB (1x MI300X from $1.85 an hour).

Published metadata for 2 models, each read from its own repository.
Field distilbert-base-uncased-finetuned-sst-2-english
distilbert/distilbert-base-uncased-finetuned-sst-2-english
distilroberta-finetuned-financial-news-sentiment-analysis
mrm8488/distilroberta-finetuned-financial-news-sentiment-analysis
Publisher DistilBERT community Manuel Romero
Task Text classification Text classification
Modality Text Text
Parameters, as reported 67M parameters 82M parameters
Architecture DistilBertForSequenceClassification RobertaForSequenceClassification
Library transformers transformers
Context length 512 tokens 514 tokens
Repository size 1.3 GB 659.8 MB
Artifact formats safetensors, onnx, pytorch, rust, tf safetensors, pytorch, tensorboard
License apache-2.0 apache-2.0
Access Open weights, no gate Open weights, no gate
Memory at 16-bit (weights and margin) 0.2 GB 0.2 GB
Cheapest GPUs at 16-bit, per hour 1x MI300X, $1.85 1x MI300X, $1.85
Memory at 4-bit (weights and margin) 0 GB 0 GB
Cheapest GPUs at 4-bit, per hour 1x MI300X, $1.85 1x MI300X, $1.85
Revision viewed 714eb0fa89d2 ae0eab9ad336
Downloads reported by the hub 3.7M 346.6k
Last observed 2026-09-18 2026-09-18

An evaluation row appears only where at least two of these models report the same benchmark with the same stated configuration, metric, unit and setup. Different evaluators stay named in each cell. Values are shown as reported: no unit conversion, no ranking.

Other Reported Results

These results are listed for each model on its own, because the conditions needed to compare them are not stated or do not match. Two results that leave a condition blank are not assumed to share it.

distilbert-base-uncased-finetuned-sst-2-english

BenchmarkConditionsResultReported byRevisionDate
glue Configuration sst2Task Text ClassificationMetric AUCComparison conditions not established 0.971663 distilbert
Publisher reported
Evaluated revision not stated
glue Configuration sst2Task Text ClassificationMetric AccuracyComparison conditions not established 0.91055 distilbert
Publisher reported
Evaluated revision not stated
glue Configuration sst2Task Text ClassificationMetric F1Comparison conditions not established 0.913717 distilbert
Publisher reported
Evaluated revision not stated
glue Configuration sst2Task Text ClassificationMetric PrecisionComparison conditions not established 0.897826 distilbert
Publisher reported
Evaluated revision not stated
glue Configuration sst2Task Text ClassificationMetric RecallComparison conditions not established 0.93018 distilbert
Publisher reported
Evaluated revision not stated
glue Configuration sst2Task Text ClassificationMetric lossComparison conditions not established 0.390139 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric AccuracyComparison conditions not established 0.988552 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric F1 MacroComparison conditions not established 0.988402 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric F1 MicroComparison conditions not established 0.988552 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric F1 WeightedComparison conditions not established 0.988555 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric Precision MacroComparison conditions not established 0.988197 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric Precision MicroComparison conditions not established 0.988552 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric Precision WeightedComparison conditions not established 0.988564 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric Recall MacroComparison conditions not established 0.988615 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric Recall MicroComparison conditions not established 0.988552 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric Recall WeightedComparison conditions not established 0.988552 distilbert
Publisher reported
Evaluated revision not stated
sst2 Configuration defaultTask Text ClassificationMetric lossComparison conditions not established 0.0406526 distilbert
Publisher reported
Evaluated revision not stated

distilroberta-finetuned-financial-news-sentiment-analysis

BenchmarkConditionsResultReported byRevisionDate
financial_phrasebank Task Text ClassificationMetric AccuracyComparison conditions not established 0.982301 mrm8488
Publisher reported
Evaluated revision not stated

SAVRN's Notes on distilbert-base-uncased-finetuned-sst-2-english

Sixty-seven million parameters, 512 tokens of context, one label per input: a classifier, not a generator, and the publisher's suggested use is topic classification. Quantizing is beside the point: 16-bit needs 0.2 GB, 8-bit 0.1 GB, the 4-bit row rounds to 0.0 GB, so run it at full precision. The cheapest Index setup is still a 192 GB MI300X at $1.85 an hour on-demand, a warehouse for a shoebox; in our facilities it shares a card with whatever else is running.

Apache 2.0 covers commercial use, modification and redistribution, notices kept, changes stated. The 512-token context means anything longer gets split upstream. It was trained on sst2 and glue, and the publisher reports 91.3 on the dev set against 92.7 for bert-base-uncased, the price of distillation. The repository holds 17 files in five formats, so pull only the one your stack loads.

Questions

Which is larger, distilbert-base-uncased-finetuned-sst-2-english or distilroberta-finetuned-financial-news-sentiment-analysis?

distilroberta-finetuned-financial-news-sentiment-analysis (82M parameters) is larger than distilbert-base-uncased-finetuned-sst-2-english (67M parameters), by the parameter counts their publishers report.

Which is cheaper to run, distilbert-base-uncased-finetuned-sst-2-english or distilroberta-finetuned-financial-news-sentiment-analysis?

At 4-bit, distilbert-base-uncased-finetuned-sst-2-english fits on 1x MI300X from $1.85 an hour and distilroberta-finetuned-financial-news-sentiment-analysis on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.

Can I use distilbert-base-uncased-finetuned-sst-2-english commercially?

Yes. distilbert-base-uncased-finetuned-sst-2-english is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

Can I use distilroberta-finetuned-financial-news-sentiment-analysis commercially?

Yes. distilroberta-finetuned-financial-news-sentiment-analysis is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

Related Comparisons