INDEX

Explanations

terms related to a geographical location, particularly the mention of "London."

oai_token-act-pair · gpt-3.5-turbo

instances of the substring "don"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

LCS

-0.72

Reviewer

-0.71

 WARNING

-0.69

ULT

-0.69

EStream

-0.65

 Belief

-0.64

 Ivory

-0.62

 Species

-0.61

 Dise

-0.61

 Healing

-0.61

POSITIVE LOGITS

't

1.00

ates

0.95

ning

0.93

ctory

0.93

ners

0.90

nell

0.89

nie

0.89

ating

0.88

etsk

0.87

ned

0.87

Activations Density 0.008%

terms related to a geographical location, particularly the mention of "London."

instances of the substring "don"

No Comments

No Known Activations