INDEX

Explanations

markers and plaques used as physical or symbolic indicators

oai_token-act-pair · gpt-3.5-turbo

terms related to indicators or symbols

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

ONSORED

-1.06

agar

-0.92

orld

-0.87

acia

-0.84

ership

-0.82

é¾

-0.82

quished

-0.79

ansk

-0.77

itals

-0.76

Kut

-0.76

POSITIVE LOGITS

 marker

1.11

 plaque

0.94

 markers

0.93

posts

0.85

 dotted

0.78

 blank

0.76

 litter

0.75

 littered

0.74

 stripe

0.74

brush

0.72

Activations Density 0.038%

markers and plaques used as physical or symbolic indicators

terms related to indicators or symbols

No Known Activations