INDEX

Explanations

words related to locations or places

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "le" 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 @ 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

raints

-0.96

 Seym

-0.96

ividual

-0.89

iaries

-0.87

ĸļ

-0.86

rador

-0.86

yrinth

-0.83

avorite

-0.83

raint

-0.82

dyl

-0.81

POSITIVE LOGITS

ttes

1.23

lla

1.15

agues

1.10

lean

0.96

isure

0.93

eping

0.88

le

0.88

rene

0.88

phant

0.87

utenant

0.83

Activations Density 0.023%

words related to locations or places

occurrences of the word "le" in various contexts

No Comments

No Known Activations