INDEX

Explanations

words related to legal proceedings like "hearing" or "confirmation hearing"

oai_token-act-pair · gpt-3.5-turbo

mentions of formal hearings or discussions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

inals

-0.84

uably

-0.66

ontent

-0.65

aughters

-0.64

vol

-0.64

odan

-0.61

aunder

-0.58

imov

-0.58

engine

-0.58

ryce

-0.55

POSITIVE LOGITS

 hearing

3.97

 Hearing

3.02

 hearings

2.17

 hear

1.83

 Hear

1.76

 listening

1.71

 hears

1.65

 seeing

1.52

 heard

1.47

 witnessing

1.39

Activations Density 0.015%

words related to legal proceedings like "hearing" or "confirmation hearing"

mentions of formal hearings or discussions

No Comments

No Known Activations