INDEX

Explanations

the word "target" with high activation values

oai_token-act-pair · gpt-3.5-turbo

words related to the concept of a "target."

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

arts

-0.83

 Amen

-0.77

 Architects

-0.72

vere

-0.70

Nich

-0.69

amas

-0.69

 Cups

-0.68

aos

-0.65

jri

-0.64

haz

-0.63

POSITIVE LOGITS

 target

3.79

 targets

3.05

target

2.74

 Target

2.14

Target

2.09

 targeted

2.06

 targeting

1.96

 targ

1.84

ARGET

1.79

 Targ

1.62

Activations Density 0.019%

the word "target" with high activation values

words related to the concept of a "target."

No Comments

No Known Activations