INDEX

Explanations

names related to a specific entity or brand, potentially a company or organization

oai_token-act-pair · gpt-3.5-turbo

references to the term "Av."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hyde

-0.67

 respons

-0.64

utenant

-0.62

 inaccessible

-0.62

 Barnett

-0.61

ptives

-0.61

 Coke

-0.60

FORMATION

-0.60

cipline

-0.60

zsche

-0.60

POSITIVE LOGITS

ocado

1.27

atars

1.19

oided

1.13

ril

1.09

atar

1.08

iol

1.03

iator

1.02

iew

1.02

iations

1.01

raham

1.00

Activations Density 0.019%

names related to a specific entity or brand, potentially a company or organization

references to the term "Av."

No Comments

No Known Activations