INDEX

Explanations

references to combo actions or items

oai_token-act-pair · gpt-3.5-turbo

references to combinations or groupings in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

mberg

-0.74

igr

-0.71

pher

-0.70

ady

-0.69

lucent

-0.68

aniel

-0.67

 Abraham

-0.65

cot

-0.65

aco

-0.64

ipple

-0.64

POSITIVE LOGITS

 combo

0.96

 combos

0.95

 Breaker

0.90

 Combo

0.86

 boxes

0.76

iques

0.71

 breaker

0.70

-+-+

0.70

 mash

0.69

-+-+-+-+

0.69

Activations Density 0.008%

references to combo actions or items

references to combinations or groupings in various contexts

No Comments

No Known Activations