INDEX

Explanations

phrases indicating a rephrasing or further explanation of content

oai_token-act-pair · gpt-3.5-turbo

phrases that introduce paraphrased or rephrased statements

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

adium

-0.68

ousel

-0.66

 heaviest

-0.66

 Columb

-0.65

 Lima

-0.64

 sund

-0.61

 lecturer

-0.60

 Clintons

-0.59

 intermediary

-0.59

 cent

-0.58

POSITIVE LOGITS

forth

0.92

hots

0.83

 guiActiveUn

0.77

phony

0.77

atility

0.77

heses

0.77

mith

0.76

tainment

0.74

ername

0.74

oire

0.73

Activations Density 0.027%

phrases indicating a rephrasing or further explanation of content

phrases that introduce paraphrased or rephrased statements

No Comments

No Known Activations