INDEX

Explanations

instances where the text indicates something is not always a certain way

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing consistency or frequency of occurrence

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

IDA

-0.85

 GENERAL

-0.79

 largeDownload

-0.76

)].

-0.70

eting

-0.69

pour

-0.67

 Flags

-0.67

ORGE

-0.66

Pages

-0.66

 Accessories

-0.65

POSITIVE LOGITS

 necessarily

0.82

 anymore

0.79

icably

0.77

 appe

0.76

 bothered

0.73

necess

0.71

 appre

0.69

 physically

0.67

 accurately

0.67

 remotely

0.67

Activations Density 0.038%

instances where the text indicates something is not always a certain way

phrases emphasizing consistency or frequency of occurrence

No Comments

No Known Activations