INDEX

Explanations

locations or landmarks, particularly mountain names

oai_token-act-pair · gpt-3.5-turbo

names of locations and geographical features

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 soph

-0.72

 literacy

-0.69

 professions

-0.67

 gamer

-0.67

 secretaries

-0.66

SAY

-0.66

 managerial

-0.65

 capacities

-0.65

 antioxid

-0.65

acters

-0.64

POSITIVE LOGITS

 Cemetery

0.98

burgh

0.94

boa

0.88

 Ridge

0.85

ushima

0.83

 Cathedral

0.82

 Landing

0.82

 Resort

0.81

 Observatory

0.80

 Island

0.80

Activations Density 0.293%

locations or landmarks, particularly mountain names

names of locations and geographical features

No Comments

No Known Activations