INDEX

Explanations

words related to opinions or comments on various topics

oai_token-act-pair · gpt-3.5-turbo

instances of objectionable or controversial topics

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

)"

-0.71

 however

-0.69

)'

-0.68

'[

-0.67

 moreover

-0.67

)</

-0.65

|--

-0.64

 meanwhile

-0.63

 depends

-0.62

*)

-0.61

POSITIVE LOGITS

boro

0.58

LOS

0.57

etime

0.56

toggle

0.54

renheit

0.53

éĹ

0.53

DCS

0.52

 gallons

0.51

Welcome

0.51

éĸ

0.50

Activations Density 2.367%

words related to opinions or comments on various topics

instances of objectionable or controversial topics

No Comments

No Known Activations