INDEX

Explanations

instances of the word "report" with varying strengths of activation

oai_token-act-pair · gpt-3.5-turbo

requests and instances of reporting various issues or concerns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

creen

-0.78

erie

-0.69

ierre

-0.62

wich

-0.62

osi

-0.60

 gauge

-0.59

frey

-0.59

hement

-0.57

aned

-0.57

ozy

-0.57

POSITIVE LOGITS

 sightings

0.77

 inacc

0.75

ribe

0.74

ufact

0.73

 misconduct

0.72

 irregularities

0.71

ounces

0.70

 inaccur

0.70

 rapes

0.69

eeds

0.68

Activations Density 0.071%

instances of the word "report" with varying strengths of activation

requests and instances of reporting various issues or concerns

No Comments

No Known Activations