INDEX

Explanations

phrases related to capability and possibility

oai_token-act-pair · gpt-3.5-turbo

phrases expressing capability or potential actions

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

Seg

-0.63

ÑĢ

-0.61

Staff

-0.60

Baltimore

-0.60

æľ

-0.59

 coincided

-0.58

Harris

-0.58

ĸļ

-0.58

Cath

-0.57

Ore

-0.57

POSITIVE LOGITS

 cheaply

0.94

 safely

0.90

 anytime

0.90

 freely

0.85

 easily

0.84

odder

0.80

 offline

0.76

 inexpensive

0.76

 effortlessly

0.75

 depending

0.75

Activations Density 0.960%

phrases related to capability and possibility

phrases expressing capability or potential actions

No Comments

No Known Activations