INDEX

Explanations

descriptions of ongoing actions or events

oai_token-act-pair · gpt-3.5-turbo

repeated instances of the phrase "going to."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

unc

-0.73

 Horus

-0.72

cher

-0.68

ifle

-0.67

ablo

-0.66

ament

-0.66

ummies

-0.65

aments

-0.64

uti

-0.63

cipl

-0.62

POSITIVE LOGITS

Ń·

0.97

 rall

0.89

 agre

0.81

 toget

0.78

 conduc

0.74

 behav

0.73

ãĥ£

0.72

ãĤĭ

0.71

 redes

0.70

©¶æ

0.69

Activations Density 0.048%

descriptions of ongoing actions or events

repeated instances of the phrase "going to."

No Comments

No Known Activations