INDEX

Explanations

mentions of the act of speaking or discussing

oai_token-act-pair · gpt-3.5-turbo

references to discussions or topics being talked about

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

 Sabha

-0.69

ogn

-0.67

gallery

-0.65

waters

-0.64

sky

-0.63

OGR

-0.61

Added

-0.61

 arsenal

-0.61

©¶æ¥µ

-0.61

quez

-0.61

POSITIVE LOGITS

ļéĨĴ

0.84

 topics

0.74

 reforming

0.73

 halfway

0.73

how

0.71

eday

0.71

nesday

0.68

 specifics

0.67

pheus

0.67

 nostalg

0.65

Activations Density 0.067%

mentions of the act of speaking or discussing

references to discussions or topics being talked about

No Comments

No Known Activations