INDEX

Explanations

formal statements of findings or judgments

oai_token-act-pair · gpt-3.5-turbo

statements and references to conclusions reached in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

timer

-0.68

Plus

-0.62

iph

-0.62

version

-0.61

unic

-0.61

 Eater

-0.59

sold

-0.59

rix

-0.59

 Plus

-0.58

phan

-0.57

POSITIVE LOGITS

 conclusions

3.67

 findings

1.76

 conclusion

1.75

 hypotheses

1.64

 judgments

1.62

 assumptions

1.61

 recommendations

1.50

 insights

1.48

 opinions

1.45

 assertions

1.45

Activations Density 0.019%

formal statements of findings or judgments

statements and references to conclusions reached in various contexts

No Comments

No Known Activations