INDEX

Explanations

references to names and titles

oai_token-act-pair · gpt-3.5-turbo

instances of the apostrophe

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

ħĭ

-0.79

ĸļ

-0.72

 eclipse

-0.69

etheless

-0.68

isphere

-0.67

nown

-0.65

icka

-0.65

swer

-0.64

sted

-0.64

 referen

-0.63

POSITIVE LOGITS

gall

0.75

esp

0.74

Allah

0.72

orange

0.72

alla

0.71

yer

0.70

morrow

0.69

hon

0.69

hair

0.69

expl

0.68

Activations Density 0.011%

references to names and titles

instances of the apostrophe

No Comments

No Known Activations