INDEX

Explanations

words related to causal relationships or connections

oai_token-act-pair · gpt-3.5-turbo

the word "that" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ron

-0.68

river

-0.67

english

-0.61

yne

-0.60

lander

-0.59

rss

-0.59

ctica

-0.59

Desk

-0.59

ner

-0.59

Kit

-0.59

POSITIVE LOGITS

 accompanies

0.79

 fateful

0.78

 preceded

0.78

 caused

0.76

 consumes

0.75

 arose

0.73

 spawned

0.73

ItemTracker

0.72

 THEY

0.71

 consumed

0.70

Activations Density 0.216%

words related to causal relationships or connections

the word "that" in various contexts

No Comments

No Known Activations