INDEX

Explanations

words related to actions that involve separation or isolation, as well as words related to exaggeration or infiltration

oai_token-act-pair · gpt-3.5-turbo

concepts related to isolation and exaggeration

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Dragonbound

-0.72

master

-0.70

 Hath

-0.69

dash

-0.69

SHIP

-0.68

 Maid

-0.68

 miscarriage

-0.68

CHAT

-0.67

dress

-0.67

beard

-0.66

POSITIVE LOGITS

ating

1.19

ates

1.18

ators

1.14

atable

1.11

isions

1.10

asion

1.09

ator

1.08

ations

1.07

ading

1.07

ocations

1.06

Activations Density 0.054%

words related to actions that involve separation or isolation, as well as words related to exaggeration or infiltration

concepts related to isolation and exaggeration

No Comments

No Known Activations