INDEX

Explanations

words related to exaggeration or overstating information

oai_token-act-pair · gpt-3.5-turbo

terms related to exaggeration or overstated claims

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

spot

-0.84

NetMessage

-0.75

aining

-0.71

abiding

-0.70

washer

-0.68

ÂłÂłÂłÂł

-0.67

tha

-0.67

avis

-0.66

cellent

-0.66

shed

-0.65

POSITIVE LOGITS

 exagger

1.25

 exaggeration

1.11

 exaggerated

0.94

 overest

0.88

 inflated

0.88

 distortions

0.81

 modesty

0.79

 guiActiveUn

0.77

0.76

 tales

0.75

Activations Density 0.021%

words related to exaggeration or overstating information

terms related to exaggeration or overstated claims

No Comments

No Known Activations