INDEX

Explanations

names of individuals or groups, possibly from news articles

oai_token-act-pair · gpt-3.5-turbo

occurrences of the special character Ļ in different contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 disadvant

-1.07

 vulner

-0.87

 mathemat

-0.86

 comprom

-0.86

 princ

-0.84

 incorpor

-0.81

 agre

-0.77

 constitu

-0.77

 fundament

-0.75

raints

-0.75

POSITIVE LOGITS

ï¸ı

1.28

ï¸

1.07

å¥³

0.93

à¥

0.92

æľ

0.86

Â½

0.84

ı

0.81

Ĩ

0.80

ãĤŃ

0.80

\":

0.79

Activations Density 0.330%

names of individuals or groups, possibly from news articles

occurrences of the special character Ļ in different contexts

No Comments

No Known Activations