INDEX

Explanations

phrases related to commands or instructions

oai_token-act-pair · gpt-3.5-turbo

punctuation marks indicating the end of sentences or significant points in text

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

kefeller

-0.75

mere

-0.69

plet

-0.67

 cures

-0.66

osaurus

-0.64

xual

-0.64

redo

-0.62

liv

-0.61

 McMaster

-0.58

iary

-0.58

POSITIVE LOGITS

Sometimes

1.06

ccording

1.03

 Suppose

0.95

Whenever

0.93

 Whenever

0.92

 Firstly

0.91

 Whilst

0.89

 Sometimes

0.88

 Notwithstanding

0.86

Unless

0.84

Activations Density 0.286%

phrases related to commands or instructions

punctuation marks indicating the end of sentences or significant points in text

No Comments

No Known Activations