INDEX

Explanations

comparative degrees and phrases indicating extremes

oai_token-act-pair · gpt-3.5-turbo

negative qualifiers and contrasting statements

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

ipeg

-0.72

 wills

-0.71

swick

-0.70

 throats

-0.67

aneers

-0.66

 clears

-0.66

 confronts

-0.65

sheets

-0.65

 Heads

-0.64

 ministries

-0.64

POSITIVE LOGITS

omorphic

0.75

omorph

0.72

TBD

0.71

 understatement

0.68

 consolation

0.66

Ļ½

0.64

actly

0.64

isible

0.62

uable

0.61

raq

0.61

Activations Density 0.783%

comparative degrees and phrases indicating extremes

negative qualifiers and contrasting statements

No Comments

No Known Activations