INDEX

Explanations

phrases related to undermining actions or concepts

oai_token-act-pair · gpt-3.5-turbo

instances of language that indicate undermining or damaging effects on credibility or arguments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

DragonMagazine

-0.72

lich

-0.67

 Cases

-0.66

 patiently

-0.64

 Stories

-0.64

saw

-0.63

ARDS

-0.62

 annotations

-0.62

anz

-0.61

 Lists

-0.61

POSITIVE LOGITS

 credibility

1.12

 fragile

1.01

 ability

0.99

 livelihood

0.97

 entire

0.93

 morale

0.88

 viability

0.88

 competitiveness

0.86

 cohesion

0.86

 usefulness

0.85

Activations Density 0.267%

phrases related to undermining actions or concepts

instances of language that indicate undermining or damaging effects on credibility or arguments

No Comments

No Known Activations