Update README to document the 2026-05-03 batch-broadcast patch (INT4 re-upload + unpatched archival).

Replace model_int4_full.onnx with batch-broadcast-patched export (Expand+Shape inserted before And); preserve unpatched at model_int4_full.unpatched.onnx for archival.

Remove stray per-tensor weight files from legacy ONNX export

Delete base.model.layers.20.self_attn.q_proj.weight with huggingface_hub

Delete base.model.layers.20.self_attn.o_proj.weight with huggingface_hub

Delete base.model.layers.20.self_attn.k_proj.weight with huggingface_hub

Delete base.model.layers.20.post_attention_layernorm.weight with huggingface_hub

Delete base.model.layers.20.mlp.up_proj.weight with huggingface_hub

Delete base.model.layers.20.mlp.gate_proj.weight with huggingface_hub

Delete base.model.layers.20.mlp.down_proj.weight with huggingface_hub

Delete base.model.layers.20.input_layernorm.weight with huggingface_hub

Delete base.model.layers.2.self_attn.v_proj.weight with huggingface_hub

Delete base.model.layers.2.self_attn.q_proj.weight with huggingface_hub

Delete base.model.layers.2.self_attn.o_proj.weight with huggingface_hub

Delete base.model.layers.2.self_attn.k_proj.weight with huggingface_hub

Delete base.model.layers.2.post_attention_layernorm.weight with huggingface_hub

Delete base.model.layers.2.mlp.up_proj.weight with huggingface_hub

Delete base.model.layers.2.mlp.gate_proj.weight with huggingface_hub

Delete base.model.layers.2.mlp.down_proj.weight with huggingface_hub

Delete base.model.layers.2.input_layernorm.weight with huggingface_hub

Delete base.model.layers.19.self_attn.v_proj.weight with huggingface_hub

Delete base.model.layers.19.self_attn.q_proj.weight with huggingface_hub

Delete base.model.layers.19.self_attn.o_proj.weight with huggingface_hub

Delete base.model.layers.19.self_attn.k_proj.weight with huggingface_hub

Delete base.model.layers.19.post_attention_layernorm.weight with huggingface_hub

Delete base.model.layers.19.mlp.up_proj.weight with huggingface_hub

Delete base.model.layers.19.mlp.gate_proj.weight with huggingface_hub

Delete base.model.layers.19.mlp.down_proj.weight with huggingface_hub

Delete base.model.layers.19.input_layernorm.weight with huggingface_hub

Delete base.model.layers.18.self_attn.v_proj.weight with huggingface_hub

Delete base.model.layers.18.self_attn.q_proj.weight with huggingface_hub

Delete base.model.layers.18.self_attn.o_proj.weight with huggingface_hub

Delete base.model.layers.18.self_attn.k_proj.weight with huggingface_hub

Delete base.model.layers.18.post_attention_layernorm.weight with huggingface_hub

Delete base.model.layers.18.mlp.up_proj.weight with huggingface_hub

Delete base.model.layers.18.mlp.gate_proj.weight with huggingface_hub

Delete base.model.layers.18.mlp.down_proj.weight with huggingface_hub

Delete base.model.layers.18.input_layernorm.weight with huggingface_hub

Delete base.model.layers.17.self_attn.v_proj.weight with huggingface_hub

Delete base.model.layers.17.self_attn.q_proj.weight with huggingface_hub

Delete base.model.layers.17.self_attn.o_proj.weight with huggingface_hub

Delete base.model.layers.17.self_attn.k_proj.weight with huggingface_hub

Delete base.model.layers.17.post_attention_layernorm.weight with huggingface_hub

Delete base.model.layers.17.mlp.up_proj.weight with huggingface_hub

Delete base.model.layers.17.mlp.gate_proj.weight with huggingface_hub

Delete base.model.layers.17.mlp.down_proj.weight with huggingface_hub

Delete base.model.layers.17.input_layernorm.weight with huggingface_hub

Delete base.model.layers.16.self_attn.v_proj.weight with huggingface_hub

Delete base.model.layers.16.self_attn.q_proj.weight with huggingface_hub

Delete base.model.layers.16.self_attn.o_proj.weight with huggingface_hub