{ "metadata": { "format": "pt", "total_size": 1304637312 }, "weight_map": { "position_embedding": "pytorch_model.bin", "token_embedding.weight": "pytorch_model.bin", "ln_f.weight": "pytorch_model.bin", "ln_f.bias": "pytorch_model.bin", "head.weight": "pytorch_model.bin", "blocks.0.ln1.weight": "pytorch_model.bin", "blocks.0.ln1.bias": "pytorch_model.bin", "blocks.0.attn.mask": "pytorch_model.bin", "blocks.0.attn.query.weight": "pytorch_model.bin", "blocks.0.attn.query.bias": "pytorch_model.bin", "blocks.0.attn.key.weight": "pytorch_model.bin", "blocks.0.attn.key.bias": "pytorch_model.bin", "blocks.0.attn.value.weight": "pytorch_model.bin", "blocks.0.attn.value.bias": "pytorch_model.bin", "blocks.0.attn.proj.weight": "pytorch_model.bin", "blocks.0.attn.proj.bias": "pytorch_model.bin", "blocks.0.ln2.weight": "pytorch_model.bin", "blocks.0.ln2.bias": "pytorch_model.bin", "blocks.0.mlp.fc1.weight": "pytorch_model.bin", "blocks.0.mlp.fc1.bias": "pytorch_model.bin", "blocks.0.mlp.fc2.weight": "pytorch_model.bin", "blocks.0.mlp.fc2.bias": "pytorch_model.bin", "blocks.1.ln1.weight": "pytorch_model.bin", "blocks.1.ln1.bias": "pytorch_model.bin", "blocks.1.attn.mask": "pytorch_model.bin", "blocks.1.attn.query.weight": "pytorch_model.bin", "blocks.1.attn.query.bias": "pytorch_model.bin", "blocks.1.attn.key.weight": "pytorch_model.bin", "blocks.1.attn.key.bias": "pytorch_model.bin", "blocks.1.attn.value.weight": "pytorch_model.bin", "blocks.1.attn.value.bias": "pytorch_model.bin", "blocks.1.attn.proj.weight": "pytorch_model.bin", "blocks.1.attn.proj.bias": "pytorch_model.bin", "blocks.1.ln2.weight": "pytorch_model.bin", "blocks.1.ln2.bias": "pytorch_model.bin", "blocks.1.mlp.fc1.weight": "pytorch_model.bin", "blocks.1.mlp.fc1.bias": "pytorch_model.bin", "blocks.1.mlp.fc2.weight": "pytorch_model.bin", "blocks.1.mlp.fc2.bias": "pytorch_model.bin", "blocks.2.ln1.weight": "pytorch_model.bin", "blocks.2.ln1.bias": "pytorch_model.bin", "blocks.2.attn.mask": "pytorch_model.bin", "blocks.2.attn.query.weight": "pytorch_model.bin", "blocks.2.attn.query.bias": "pytorch_model.bin", "blocks.2.attn.key.weight": "pytorch_model.bin", "blocks.2.attn.key.bias": "pytorch_model.bin", "blocks.2.attn.value.weight": "pytorch_model.bin", "blocks.2.attn.value.bias": "pytorch_model.bin", "blocks.2.attn.proj.weight": "pytorch_model.bin", "blocks.2.attn.proj.bias": "pytorch_model.bin", "blocks.2.ln2.weight": "pytorch_model.bin", "blocks.2.ln2.bias": "pytorch_model.bin", "blocks.2.mlp.fc1.weight": "pytorch_model.bin", "blocks.2.mlp.fc1.bias": "pytorch_model.bin", "blocks.2.mlp.fc2.weight": "pytorch_model.bin", "blocks.2.mlp.fc2.bias": "pytorch_model.bin", "blocks.3.ln1.weight": "pytorch_model.bin", "blocks.3.ln1.bias": "pytorch_model.bin", "blocks.3.attn.mask": "pytorch_model.bin", "blocks.3.attn.query.weight": "pytorch_model.bin", "blocks.3.attn.query.bias": "pytorch_model.bin", "blocks.3.attn.key.weight": "pytorch_model.bin", "blocks.3.attn.key.bias": "pytorch_model.bin", "blocks.3.attn.value.weight": "pytorch_model.bin", "blocks.3.attn.value.bias": "pytorch_model.bin", "blocks.3.attn.proj.weight": "pytorch_model.bin", "blocks.3.attn.proj.bias": "pytorch_model.bin", "blocks.3.ln2.weight": "pytorch_model.bin", "blocks.3.ln2.bias": "pytorch_model.bin", "blocks.3.mlp.fc1.weight": "pytorch_model.bin", "blocks.3.mlp.fc1.bias": "pytorch_model.bin", "blocks.3.mlp.fc2.weight": "pytorch_model.bin", "blocks.3.mlp.fc2.bias": "pytorch_model.bin", "blocks.4.ln1.weight": "pytorch_model.bin", "blocks.4.ln1.bias": "pytorch_model.bin", "blocks.4.attn.mask": "pytorch_model.bin", "blocks.4.attn.query.weight": "pytorch_model.bin", "blocks.4.attn.query.bias": "pytorch_model.bin", "blocks.4.attn.key.weight": "pytorch_model.bin", "blocks.4.attn.key.bias": "pytorch_model.bin", "blocks.4.attn.value.weight": "pytorch_model.bin", "blocks.4.attn.value.bias": "pytorch_model.bin", "blocks.4.attn.proj.weight": "pytorch_model.bin", "blocks.4.attn.proj.bias": "pytorch_model.bin", "blocks.4.ln2.weight": "pytorch_model.bin", "blocks.4.ln2.bias": "pytorch_model.bin", "blocks.4.mlp.fc1.weight": "pytorch_model.bin", "blocks.4.mlp.fc1.bias": "pytorch_model.bin", "blocks.4.mlp.fc2.weight": "pytorch_model.bin", "blocks.4.mlp.fc2.bias": "pytorch_model.bin", "blocks.5.ln1.weight": "pytorch_model.bin", "blocks.5.ln1.bias": "pytorch_model.bin", "blocks.5.attn.mask": "pytorch_model.bin", "blocks.5.attn.query.weight": "pytorch_model.bin", "blocks.5.attn.query.bias": "pytorch_model.bin", "blocks.5.attn.key.weight": "pytorch_model.bin", "blocks.5.attn.key.bias": "pytorch_model.bin", "blocks.5.attn.value.weight": "pytorch_model.bin", "blocks.5.attn.value.bias": "pytorch_model.bin", "blocks.5.attn.proj.weight": "pytorch_model.bin", "blocks.5.attn.proj.bias": "pytorch_model.bin", "blocks.5.ln2.weight": "pytorch_model.bin", "blocks.5.ln2.bias": "pytorch_model.bin", "blocks.5.mlp.fc1.weight": "pytorch_model.bin", "blocks.5.mlp.fc1.bias": "pytorch_model.bin", "blocks.5.mlp.fc2.weight": "pytorch_model.bin", "blocks.5.mlp.fc2.bias": "pytorch_model.bin", "blocks.6.ln1.weight": "pytorch_model.bin", "blocks.6.ln1.bias": "pytorch_model.bin", "blocks.6.attn.mask": "pytorch_model.bin", "blocks.6.attn.query.weight": "pytorch_model.bin", "blocks.6.attn.query.bias": "pytorch_model.bin", "blocks.6.attn.key.weight": "pytorch_model.bin", "blocks.6.attn.key.bias": "pytorch_model.bin", "blocks.6.attn.value.weight": "pytorch_model.bin", "blocks.6.attn.value.bias": "pytorch_model.bin", "blocks.6.attn.proj.weight": "pytorch_model.bin", "blocks.6.attn.proj.bias": "pytorch_model.bin", "blocks.6.ln2.weight": "pytorch_model.bin", "blocks.6.ln2.bias": "pytorch_model.bin", "blocks.6.mlp.fc1.weight": "pytorch_model.bin", "blocks.6.mlp.fc1.bias": "pytorch_model.bin", "blocks.6.mlp.fc2.weight": "pytorch_model.bin", "blocks.6.mlp.fc2.bias": "pytorch_model.bin", "blocks.7.ln1.weight": "pytorch_model.bin", "blocks.7.ln1.bias": "pytorch_model.bin", "blocks.7.attn.mask": "pytorch_model.bin", "blocks.7.attn.query.weight": "pytorch_model.bin", "blocks.7.attn.query.bias": "pytorch_model.bin", "blocks.7.attn.key.weight": "pytorch_model.bin", "blocks.7.attn.key.bias": "pytorch_model.bin", "blocks.7.attn.value.weight": "pytorch_model.bin", "blocks.7.attn.value.bias": "pytorch_model.bin", "blocks.7.attn.proj.weight": "pytorch_model.bin", "blocks.7.attn.proj.bias": "pytorch_model.bin", "blocks.7.ln2.weight": "pytorch_model.bin", "blocks.7.ln2.bias": "pytorch_model.bin", "blocks.7.mlp.fc1.weight": "pytorch_model.bin", "blocks.7.mlp.fc1.bias": "pytorch_model.bin", "blocks.7.mlp.fc2.weight": "pytorch_model.bin", "blocks.7.mlp.fc2.bias": "pytorch_model.bin", "blocks.8.ln1.weight": "pytorch_model.bin", "blocks.8.ln1.bias": "pytorch_model.bin", "blocks.8.attn.mask": "pytorch_model.bin", "blocks.8.attn.query.weight": "pytorch_model.bin", "blocks.8.attn.query.bias": "pytorch_model.bin", "blocks.8.attn.key.weight": "pytorch_model.bin", "blocks.8.attn.key.bias": "pytorch_model.bin", "blocks.8.attn.value.weight": "pytorch_model.bin", "blocks.8.attn.value.bias": "pytorch_model.bin", "blocks.8.attn.proj.weight": "pytorch_model.bin", "blocks.8.attn.proj.bias": "pytorch_model.bin", "blocks.8.ln2.weight": "pytorch_model.bin", "blocks.8.ln2.bias": "pytorch_model.bin", "blocks.8.mlp.fc1.weight": "pytorch_model.bin", "blocks.8.mlp.fc1.bias": "pytorch_model.bin", "blocks.8.mlp.fc2.weight": "pytorch_model.bin", "blocks.8.mlp.fc2.bias": "pytorch_model.bin", "blocks.9.ln1.weight": "pytorch_model.bin", "blocks.9.ln1.bias": "pytorch_model.bin", "blocks.9.attn.mask": "pytorch_model.bin", "blocks.9.attn.query.weight": "pytorch_model.bin", "blocks.9.attn.query.bias": "pytorch_model.bin", "blocks.9.attn.key.weight": "pytorch_model.bin", "blocks.9.attn.key.bias": "pytorch_model.bin", "blocks.9.attn.value.weight": "pytorch_model.bin", "blocks.9.attn.value.bias": "pytorch_model.bin", "blocks.9.attn.proj.weight": "pytorch_model.bin", "blocks.9.attn.proj.bias": "pytorch_model.bin", "blocks.9.ln2.weight": "pytorch_model.bin", "blocks.9.ln2.bias": "pytorch_model.bin", "blocks.9.mlp.fc1.weight": "pytorch_model.bin", "blocks.9.mlp.fc1.bias": "pytorch_model.bin", "blocks.9.mlp.fc2.weight": "pytorch_model.bin", "blocks.9.mlp.fc2.bias": "pytorch_model.bin", "blocks.10.ln1.weight": "pytorch_model.bin", "blocks.10.ln1.bias": "pytorch_model.bin", "blocks.10.attn.mask": "pytorch_model.bin", "blocks.10.attn.query.weight": "pytorch_model.bin", "blocks.10.attn.query.bias": "pytorch_model.bin", "blocks.10.attn.key.weight": "pytorch_model.bin", "blocks.10.attn.key.bias": "pytorch_model.bin", "blocks.10.attn.value.weight": "pytorch_model.bin", "blocks.10.attn.value.bias": "pytorch_model.bin", "blocks.10.attn.proj.weight": "pytorch_model.bin", "blocks.10.attn.proj.bias": "pytorch_model.bin", "blocks.10.ln2.weight": "pytorch_model.bin", "blocks.10.ln2.bias": "pytorch_model.bin", "blocks.10.mlp.fc1.weight": "pytorch_model.bin", "blocks.10.mlp.fc1.bias": "pytorch_model.bin", "blocks.10.mlp.fc2.weight": "pytorch_model.bin", "blocks.10.mlp.fc2.bias": "pytorch_model.bin", "blocks.11.ln1.weight": "pytorch_model.bin", "blocks.11.ln1.bias": "pytorch_model.bin", "blocks.11.attn.mask": "pytorch_model.bin", "blocks.11.attn.query.weight": "pytorch_model.bin", "blocks.11.attn.query.bias": "pytorch_model.bin", "blocks.11.attn.key.weight": "pytorch_model.bin", "blocks.11.attn.key.bias": "pytorch_model.bin", "blocks.11.attn.value.weight": "pytorch_model.bin", "blocks.11.attn.value.bias": "pytorch_model.bin", "blocks.11.attn.proj.weight": "pytorch_model.bin", "blocks.11.attn.proj.bias": "pytorch_model.bin", "blocks.11.ln2.weight": "pytorch_model.bin", "blocks.11.ln2.bias": "pytorch_model.bin", "blocks.11.mlp.fc1.weight": "pytorch_model.bin", "blocks.11.mlp.fc1.bias": "pytorch_model.bin", "blocks.11.mlp.fc2.weight": "pytorch_model.bin", "blocks.11.mlp.fc2.bias": "pytorch_model.bin" } }