INDEX

Explanations

references to authors or writers and their respective publications

oai_token-act-pair · gpt-3.5-turbo

sentences that indicate professional roles or identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 stamped

-0.77

onga

-0.76

 extinguished

-0.76

 trap

-0.68

 toast

-0.67

 orderly

-0.67

 brut

-0.67

 chopping

-0.66

 alley

-0.66

aceae

-0.63

POSITIVE LOGITS

 Previously

1.13

 Follow

0.97

 Email

0.93

 Readers

0.92

 Currently

0.90

Its

0.90

 Originally

0.89

He

0.87

 Visit

0.86

His

0.85

Activations Density 0.167%

references to authors or writers and their respective publications

sentences that indicate professional roles or identities

No Comments

No Known Activations