INDEX

Explanations

positive adjectives such as "impressive," "strong," "stellar," "lengthy," "invaluable," "generous," "dismal," "incredible," "bizarre," "eerie," "explosive," "surprising," "horrendous," "

oai_token-act-pair · gpt-3.5-turbo

adjectives indicating extreme qualities or characteristics

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

uther

-0.76

AAAA

-0.72

Gi

-0.69

ecause

-0.69

 Rohing

-0.69

vertising

-0.69

rarily

-0.69

BU

-0.69

Specific

-0.68

 Yourself

-0.68

POSITIVE LOGITS

 fortunes

1.02

 foray

1.00

 journey

0.97

 beginnings

0.93

 counterparts

0.92

 arsenal

0.91

 creations

0.90

 exploits

0.90

 predecessors

0.90

 antics

0.88

Activations Density 0.230%

positive adjectives such as "impressive," "strong," "stellar," "lengthy," "invaluable," "generous," "dismal," "incredible," "bizarre," "eerie," "explosive," "surprising," "horrendous," "

adjectives indicating extreme qualities or characteristics

No Comments

No Known Activations