INDEX

Explanations

trigger words signaling a transition or change in topic

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Before."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 pers

-0.72

asc

-0.70

 perpet

-0.65

 fest

-0.60

 unspecified

-0.59

 trail

-0.59

 cent

-0.58

swe

-0.58

 Noble

-0.57

 reincarn

-0.57

POSITIVE LOGITS

Before

3.09

 Before

2.10

before

1.90

After

1.78

Prior

1.64

During

1.64

Without

1.64

Until

1.60

Previously

1.52

Since

1.52

Activations Density 0.020%

trigger words signaling a transition or change in topic

instances of the word "Before."

No Comments

No Known Activations