Features upgrade of Embedding model (quic#424)

1. Added `Pooling` support to Embedding Model
2. Added multiple seq_len support for the embedding model using
`QEffAutoModel`
4. Added test for pooling and multiple seq_len

---------

Signed-off-by: Amit Raj <quic_amitraj@quicinc.com>
Signed-off-by: Abukhoyer Shaik <abukhoye@qti.qualcomm.com>
Co-authored-by: Abukhoyer Shaik <abukhoye@qti.qualcomm.com>

Author: 2 people

QEfficient/transformers/models/pytorch_transforms.py

@@ -144,6 +144,7 @@
 from QEfficient.customop import CustomRMSNormAIC, GemmaCustomRMSNormAIC
 from QEfficient.transformers.embeddings.embedding_utils import POOLING_MAP, PooledModel, validate_user_pooling_function
 from QEfficient.transformers.cache_utils import QEffDynamicCache
+from QEfficient.transformers.embeddings.embedding_utils import POOLING_MAP, PooledModel, validate_user_pooling_function
 from QEfficient.transformers.models.codegen.modeling_codegen import (
     QEffCodeGenAttention,
     QeffCodeGenBlock,