INDEX

Explanations

general terms or concepts

oai_token-act-pair · gpt-3.5-turbo

statements regarding generalizations about groups or categories of people and situations

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

itial

-0.66

¥

-0.66

letes

-0.64

gged

-0.63

 Called

-0.63

bang

-0.62

 Prin

-0.62

lete

-0.61

line

-0.59

poke

-0.58

POSITIVE LOGITS

 relativity

0.95

 terms

0.86

edia

0.76

orea

0.68

 è£ıè

0.66

 ILCS

0.65

terms

0.64

ombat

0.64

dayName

0.62

eteria

0.62

Activations Density 0.050%

general terms or concepts

statements regarding generalizations about groups or categories of people and situations

No Comments

No Known Activations