INDEX

Explanations

words related to reaching a certain level or achievement

oai_token-act-pair · gpt-3.5-turbo

instances of the word "reaches."

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

Negative Logits

ressed

-0.67

cussion

-0.60

ston

-0.59

uty

-0.57

Lot

-0.56

 shame

-0.56

 Mock

-0.55

inese

-0.55

error

-0.54

ition

-0.54

POSITIVE LOGITS

 reaches

3.46

 reach

1.88

 reached

1.79

 reaching

1.58

 exceeds

1.43

 achieves

1.43

 arrives

1.28

 rises

1.26

 enters

1.26

 stretches

1.24

Activations Density 0.010%

words related to reaching a certain level or achievement

instances of the word "reaches."

No Known Activations