INDEX

Explanations

numeric comparisons such as "more than" or "less than"

oai_token-act-pair · gpt-3.5-turbo

comparative expressions involving quantities or numerical limits

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

onder

-0.67

orers

-0.65

roller

-0.63

 Dynamics

-0.63

 Eliot

-0.60

NEY

-0.59

Robin

-0.58

olitan

-0.58

PDATE

-0.57

orde

-0.57

POSITIVE LOGITS

 usual

0.82

onyms

0.80

idth

0.75

daq

0.74

normal

0.73

capacity

0.72

0.70

average

0.69

 specified

0.69

 allowable

0.68

Activations Density 0.094%

numeric comparisons such as "more than" or "less than"

comparative expressions involving quantities or numerical limits

No Comments

No Known Activations