INDEX

Explanations

numbers indicating a decrease or downscaling in various contexts

oai_token-act-pair · gpt-3.5-turbo

phrases indicating decreases or declines in various contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

-0.71

ature

-0.70

aceous

-0.68

 Pros

-0.67

ista

-0.62

ãģ®éŃĶ

-0.61

 Portal

-0.59

 Gears

-0.58

eering

-0.57

POSITIVE LOGITS

graded

1.32

grading

1.18

stairs

1.14

LOAD

1.04

grades

1.00

hill

0.95

linked

0.93

played

0.92

pour

0.90

loaded

0.90

Activations Density 0.035%

numbers indicating a decrease or downscaling in various contexts

phrases indicating decreases or declines in various contexts

No Comments

No Known Activations