INDEX

Explanations

questions related to interests or preferences

oai_token-act-pair · gpt-3.5-turbo

phrases related to emotional reactions or responses

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

.).

-0.64

llular

-0.61

 overtime

-0.61

sic

-0.60

)."

-0.60

net

-0.59

dyl

-0.59

 aven

-0.59

.</

-0.59

omever

-0.58

POSITIVE LOGITS

Own

0.97

 Relationship

0.90

 Started

0.89

 Loop

0.87

 Clause

0.84

 Behind

0.84

 Affect

0.83

 Case

0.82

 Approach

0.82

 Guys

0.81

Activations Density 0.558%

questions related to interests or preferences

phrases related to emotional reactions or responses

No Comments

No Known Activations