INDEX

Explanations

words related to things thought to be undeniable or unquestionable facts

oai_token-act-pair · gpt-3.5-turbo

occurrences of the string "ir" 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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Concord

-0.74

 Crew

-0.74

soDeliveryDate

-0.74

 Parad

-0.72

olitan

-0.71

hetti

-0.70

geist

-0.67

uyomi

-0.67

hower

-0.66

rooms

-0.66

POSITIVE LOGITS

ir

1.06

cles

0.88

ked

0.87

uably

0.79

bestos

0.75

 fortun

0.74

idium

0.74

lessly

0.73

 destro

0.73

anthrop

0.69

Activations Density 0.005%

words related to things thought to be undeniable or unquestionable facts

occurrences of the string "ir" in various contexts

No Comments

No Known Activations