|
{ |
|
"encoder.block.0.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.0.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.0.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.0.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.0.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.0.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.0.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.1.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.1.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.1.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.1.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.1.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.1.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.1.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.2.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.2.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.2.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.2.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.2.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.2.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.2.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.3.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.3.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.3.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.3.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.3.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.3.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.3.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.4.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.4.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.4.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.4.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.4.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.4.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.4.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.5.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.5.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.5.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.5.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.5.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.5.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.5.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.6.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.6.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.6.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.6.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.6.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.6.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.6.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.7.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.7.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.7.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.7.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.7.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.7.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.7.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.8.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.8.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.8.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.8.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.8.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.8.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.8.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.9.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.9.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.9.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.9.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.9.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.9.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.9.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.10.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.10.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.10.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.10.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.10.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.10.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.10.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.11.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.11.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.11.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.11.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.11.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.11.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.11.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.12.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.12.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.12.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.12.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.12.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.12.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.12.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.13.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.13.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.13.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.13.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.13.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.13.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.13.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.14.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.14.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.14.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.14.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.14.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.14.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.14.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.15.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.15.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.15.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.15.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.15.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.15.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.15.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.16.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.16.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.16.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.16.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.16.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.16.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.16.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.17.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.17.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.17.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.17.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.17.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.17.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.17.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.18.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.18.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.18.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.18.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.18.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.18.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.18.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.19.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.19.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.19.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.19.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.19.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.19.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.19.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.20.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.20.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.20.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.20.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.20.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.20.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.20.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.21.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.21.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.21.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.21.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.21.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.21.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.21.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.22.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.22.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.22.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.22.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.22.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.22.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.22.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.23.layer.0.SelfAttention.q": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.23.layer.0.SelfAttention.k": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.23.layer.0.SelfAttention.v": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.23.layer.0.SelfAttention.o": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.23.layer.1.DenseReluDense.wi_0": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.23.layer.1.DenseReluDense.wi_1": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
}, |
|
"encoder.block.23.layer.1.DenseReluDense.wo": { |
|
"weights": "qfloat8_e4m3fn", |
|
"activations": "none" |
|
} |
|
} |