INDEX

Explanations

terms indicating a preference for one option over another

oai_token-act-pair · gpt-3.5-turbo

statements expressing preference or choice

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

breaks

-0.79

runner

-0.74

Runner

-0.73

esi

-0.67

orig

-0.65

eval

-0.65

INAL

-0.64

uss

-0.62

 runner

-0.61

ults

-0.61

POSITIVE LOGITS

 than

0.81

 tolerate

0.74

":["

0.71

 prioritize

0.71

 Than

0.69

 accommodate

0.68

otomy

0.66

 afford

0.65

 cater

0.64

 Intelligent

0.64

Activations Density 0.015%

terms indicating a preference for one option over another

statements expressing preference or choice

No Comments

No Known Activations