INDEX

Explanations

single letters followed by words starting with "im"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "im."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

NetMessage

-0.97

chnology

-0.77

 Parenthood

-0.69

ĸļ

-0.69

 Nicarag

-0.69

 asbestos

-0.67

¬¼

-0.66

ĪĴ

-0.65

 grease

-0.64

Ĥª

-0.64

POSITIVE LOGITS

mediately

1.24

agine

1.18

ple

1.18

perial

1.12

otor

1.10

enez

1.09

ension

1.08

ulus

1.08

plement

1.07

ortal

1.06

Activations Density 0.032%

single letters followed by words starting with "im"

occurrences of the word "im."

No Comments

No Known Activations