INDEX

Explanations

words related to physical objects that can be touched or perceived in a concrete way

oai_token-act-pair · gpt-3.5-turbo

references to tangible versus intangible concepts

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

ammers

-0.83

oly

-0.76

kers

-0.73

ften

-0.73

enhagen

-0.73

annis

-0.72

nen

-0.70

missions

-0.69

 rampant

-0.67

doctor

-0.67

POSITIVE LOGITS

 tangible

0.89

 manifestation

0.85

atform

0.85

isable

0.81

izable

0.81

TY

0.81

itarian

0.76

angible

0.74

 concrete

0.73

çĲ

0.72

Activations Density 0.024%

words related to physical objects that can be touched or perceived in a concrete way

references to tangible versus intangible concepts

No Comments

No Known Activations