INDEX

Explanations

locations or places

oai_token-act-pair · gpt-3.5-turbo

the word "near" and its variations often indicating proximity

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Upload

-0.83

ilus

-0.81

Mit

-0.80

STATE

-0.75

ãĤ¸

-0.73

ICAN

-0.72

UL

-0.72

uese

-0.72

ilts

-0.71

glers

-0.70

POSITIVE LOGITS

sighted

1.04

shore

0.84

 vicinity

0.80

 misses

0.79

 proximity

0.78

ctic

0.73

 mint

0.71

 near

0.71

 resemblance

0.70

entimes

0.70

Activations Density 0.030%

locations or places

the word "near" and its variations often indicating proximity

No Comments

No Known Activations