INDEX

Explanations

words related to criticisms or negative evaluations

oai_token-act-pair · gpt-3.5-turbo

instances of the prefix "bl" followed by various suffixes or forms

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

HER

-0.80

Construct

-0.75

UAL

-0.70

ETF

-0.70

 Emir

-0.69

geist

-0.68

RAL

-0.68

pour

-0.68

ADRA

-0.68

heid

-0.67

POSITIVE LOGITS

ossom

1.12

anche

1.09

anches

1.08

ameless

1.07

ogging

1.04

izzard

1.01

adders

1.00

ocking

0.99

itter

0.98

uff

0.97

Activations Density 0.010%

words related to criticisms or negative evaluations

instances of the prefix "bl" followed by various suffixes or forms

No Comments

No Known Activations