INDEX

Explanations

references to geographical locations or points of interest

oai_token-act-pair · gpt-3.5-turbo

references to specific locations or landmarks named "Point."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Interstitial

-0.88

uthor

-0.82

natureconservancy

-0.80

ESE

-0.80

 delinqu

-0.77

ÄŁ

-0.77

ÃįÃį

-0.76

versely

-0.70

ourt

-0.69

XY

-0.68

POSITIVE LOGITS

point

0.98

hole

0.94

lessly

0.88

points

0.81

 Point

0.80

itude

0.80

hai

0.77

patrick

0.75

edly

0.75

holes

0.75

Activations Density 0.031%

references to geographical locations or points of interest

references to specific locations or landmarks named "Point."

No Comments

No Known Activations