INDEX

Explanations

terms related to specific styles or formats

oai_token-act-pair · gpt-3.5-turbo

references to Soviet-style elements or influences

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

umi

-0.84

ora

-0.82

ayson

-0.79

unes

-0.78

wright

-0.74

å¥³

-0.74

oras

-0.73

inosaur

-0.73

yrus

-0.73

iband

-0.73

POSITIVE LOGITS

lihood

0.94

 executions

0.77

 style

0.71

 immersion

0.69

 consistency

0.68

 impunity

0.68

 offenses

0.67

style

0.66

 Dragonbound

0.65

 inference

0.63

Activations Density 0.017%

terms related to specific styles or formats

references to Soviet-style elements or influences

No Comments

No Known Activations