INDEX

Explanations

state names

oai_token-act-pair · gpt-3.5-turbo

references to governmental or political entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

swe

-0.72

 mant

-0.69

 pudding

-0.66

 figur

-0.65

 mould

-0.65

 newcomers

-0.64

 heels

-0.64

 arom

-0.64

 Argent

-0.64

 narciss

-0.63

POSITIVE LOGITS

State

3.91

state

2.68

STATE

2.50

States

2.46

 State

2.44

 STATE

2.40

 state

2.05

states

1.91

 States

1.46

Gov

1.44

Activations Density 0.010%

state names

references to governmental or political entities

No Comments

No Known Activations