INDEX

Explanations

instances where something or someone is the first to achieve a particular milestone or accomplishment

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate "first" events or achievements

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

quickShipAvailable

-0.74

tun

-0.69

Services

-0.66

Fine

-0.66

jon

-0.65

 --------------------------------

-0.65

cas

-0.63

 Directions

-0.63

 Continued

-0.62

Dat

-0.62

POSITIVE LOGITS

pless

0.97

ggles

0.92

wered

0.91

 congratulate

0.89

 receive

0.89

 ever

0.89

 reach

0.88

 achieve

0.88

 inspire

0.87

 liberate

0.87

Activations Density 0.072%

instances where something or someone is the first to achieve a particular milestone or accomplishment

phrases that indicate "first" events or achievements

No Comments

No Known Activations