INDEX

Explanations

dates and years

oai_token-act-pair · gpt-3.5-turbo

specific years and dates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 clipboard

-0.62

ndra

-0.61

seed

-0.58

edge

-0.56

wered

-0.56

 disemb

-0.56

 hungry

-0.55

 subreddit

-0.54

ipple

-0.54

Edge

-0.54

POSITIVE LOGITS

-'

0.93

å¹

0.85

ãĥŁ

0.81

ĸļ

0.75

 onwards

0.72

 ï¿½ï¿½ï¿½ï¿½

0.68

ãĤ¦ãĤ¹

0.66

 onward

0.65

BCE

0.65

âĢ²

0.64

Activations Density 0.072%

dates and years

specific years and dates

No Comments

No Known Activations