INDEX

Explanations

words related to suspicion or suspicion-like concepts

oai_token-act-pair · gpt-3.5-turbo

terms related to suspicion and the action of suspending

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Kaiser

-0.68

 Cour

-0.67

WER

-0.66

eer

-0.65

ORGE

-0.63

IFIED

-0.62

 baker

-0.61

atered

-0.61

PRES

-0.61

 trib

-0.60

POSITIVE LOGITS

susp

1.29

ension

1.18

usp

1.12

 suspend

1.08

ensions

1.08

 Susp

1.07

olicy

0.97

icious

0.92

ended

0.90

 suspensions

0.87

Activations Density 0.006%

words related to suspicion or suspicion-like concepts

terms related to suspicion and the action of suspending

No Comments

No Known Activations