INDEX

Explanations

phrases related to starting or engaging in a conversation

oai_token-act-pair · gpt-3.5-turbo

mentions of conversations

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

rule

-0.72

cheat

-0.70

arah

-0.65

rikes

-0.65

anmar

-0.64

emale

-0.64

metics

-0.62

atever

-0.62

redit

-0.61

peria

-0.60

POSITIVE LOGITS

 conversation

0.93

 conversations

0.87

ogue

0.86

 banter

0.83

naire

0.80

 Conversation

0.77

BACK

0.77

 overheard

0.76

 invitations

0.75

 transcripts

0.71

Activations Density 0.022%

phrases related to starting or engaging in a conversation

mentions of conversations

No Comments

No Known Activations