INDEX

Explanations

actions or decisions taken instead of others

oai_token-act-pair · gpt-3.5-turbo

gerunds and phrases suggesting actions or processes

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

okÃ©

-0.62

 congratulated

-0.62

 Nanto

-0.62

soon

-0.60

ista

-0.59

anian

-0.59

stad

-0.58

ASA

-0.57

âĹ¼

-0.56

 kindred

-0.55

POSITIVE LOGITS

 altogether

0.91

 anymore

0.90

 anything

0.79

 outright

0.78

ĨĴ

0.71

 necessarily

0.70

nor

0.69

traditional

0.67

 actual

0.66

 versa

0.65

Activations Density 0.188%

actions or decisions taken instead of others

gerunds and phrases suggesting actions or processes

No Comments

No Known Activations