INDEX

Explanations

words related to specific individuals or figures

oai_token-act-pair · gpt-3.5-turbo

words with specific suffixes related to technical or formal 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

 resonance

-0.68

 Reporter

-0.65

 watershed

-0.62

 polls

-0.60

 soak

-0.59

 delay

-0.58

 rumors

-0.57

 backlog

-0.57

 loophole

-0.56

 audi

-0.55

POSITIVE LOGITS

lict

0.87

iture

0.76

initialized

0.75

monary

0.75

licts

0.72

ret

0.71

omics

0.71

ript

0.71

ography

0.71

mares

0.71

Activations Density 0.323%

words related to specific individuals or figures

words with specific suffixes related to technical or formal concepts

No Comments

No Known Activations