INDEX

Explanations

occurrences of phrases related to exposure of information or secrets

oai_token-act-pair · gpt-3.5-turbo

variations of the verb "blow," particularly in contexts related to revealing or disclosing information

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

ournal

-0.70

ç«

-0.66

avis

-0.65

 Clause

-0.64

lov

-0.63

ript

-0.63

States

-0.62

 Gael

-0.62

Lair

-0.60

Definition

-0.60

POSITIVE LOGITS

guns

0.91

 holes

0.82

out

0.82

 whist

0.82

 blown

0.81

outs

0.79

 away

0.77

blown

0.76

gun

0.76

pipe

0.75

Activations Density 0.012%

occurrences of phrases related to exposure of information or secrets

variations of the verb "blow," particularly in contexts related to revealing or disclosing information

No Comments

No Known Activations