INDEX

Explanations

action or intention-triggering phrases

oai_token-act-pair · gpt-3.5-turbo

punctuation marks or symbols typically indicating pauses or interruptions in text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ãĥ¼ãĥĨãĤ£

-0.92

nen

-0.84

ãĥ¼ãĥĨ

-0.80

onom

-0.70

ãĥİ

-0.70

isers

-0.70

iled

-0.68

izers

-0.67

leneck

-0.67

emia

-0.66

POSITIVE LOGITS

picture

0.82

LOG

0.80

jpg

0.80

etc

0.79

hammad

0.79

...]

0.78

mmmm

0.77

pict

0.77

sym

0.75

CONT

0.75

Activations Density 0.011%

action or intention-triggering phrases

punctuation marks or symbols typically indicating pauses or interruptions in text

No Comments

No Known Activations