INDEX

Explanations

words related to the TOR (The Onion Router) network, potentially indicating a focus on technology or security

oai_token-act-pair · gpt-3.5-turbo

mentions of "Toronto" and references to the Syndicate

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 kinderg

-0.72

 balls

-0.68

 Blank

-0.65

 Dean

-0.64

 Balls

-0.63

 halls

-0.63

 Wolf

-0.63

 dummy

-0.62

Fe

-0.62

 college

-0.61

POSITIVE LOGITS

TOR

4.74

TOR

2.77

Toronto

1.35

Tor

1.24

tor

1.12

TOP

1.09

SAN

1.04

LOS

1.01

POR

1.01

Ont

1.01

Activations Density 0.016%

words related to the TOR (The Onion Router) network, potentially indicating a focus on technology or security

mentions of "Toronto" and references to the Syndicate

No Comments

No Known Activations