INDEX

Explanations

names of individuals or characters

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "llo" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

stood

-0.75

CPC

-0.69

regular

-0.69

lished

-0.69

ãĥĩ

-0.66

ERAL

-0.65

ORT

-0.64

inals

-0.63

ORN

-0.63

Rare

-0.63

POSITIVE LOGITS

llo

1.25

zzi

1.01

zzle

0.89

zzo

0.83

abba

0.79

otti

0.78

osta

0.76

edo

0.74

igan

0.72

etti

0.71

Activations Density 0.001%

names of individuals or characters

occurrences of the word "llo" in various contexts

No Comments

No Known Activations