INDEX

Explanations

mentions of things being connected or associated with each other

oai_token-act-pair · gpt-3.5-turbo

references to the concept of being "tied" 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 @ 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

ASE

-0.84

å§«

-0.78

afa

-0.75

XY

-0.69

 PLAN

-0.69

ç¥ŀ

-0.67

aya

-0.65

hma

-0.64

ista

-0.62

 Galaxy

-0.62

POSITIVE LOGITS

 tied

1.35

 tying

1.10

tie

0.97

tie

0.91

nesday

0.91

quartered

0.84

 knot

0.83

pipe

0.83

bare

0.82

uctor

0.80

Activations Density 0.012%

mentions of things being connected or associated with each other

references to the concept of being "tied" in various contexts

No Comments

No Known Activations