minima / minima_config.json
ProCreations's picture
Pin upstream LFM revision
b05e5b8 verified
Raw
History Blame Contribute Delete
18.2 kB
{
"format_version": 1,
"format": "i2_s",
"logical_weight_bits": 1.585,
"physical_weight_bits": 2,
"activation_bits": 8,
"base_model": "LiquidAI/LFM2.5-Encoder-350M",
"model_kind": "encoder",
"max_context": 8192,
"modules": {
"embed_tokens": {
"kind": "embedding",
"num_embeddings": 65536,
"embedding_dim": 1024,
"padding_idx": 0,
"group_size": 32,
"recovery_rank": 128
},
"layers.0.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.0.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.0.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.0.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.0.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.1.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.1.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.1.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.1.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.1.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.2.self_attn.q_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.2.self_attn.k_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.2.self_attn.v_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.2.self_attn.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.2.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.2.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.2.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.3.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.3.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.3.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.3.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.3.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.4.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.4.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.4.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.4.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.4.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.5.self_attn.q_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.5.self_attn.k_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.5.self_attn.v_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.5.self_attn.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.5.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.5.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.5.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.6.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.6.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.6.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.6.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.6.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.7.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.7.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.7.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.7.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.7.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.8.self_attn.q_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.8.self_attn.k_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.8.self_attn.v_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.8.self_attn.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.8.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.8.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.8.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.9.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.9.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.9.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.9.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.9.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.10.self_attn.q_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.10.self_attn.k_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.10.self_attn.v_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.10.self_attn.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.10.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.10.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.10.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.11.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.11.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.11.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.11.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.11.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.12.self_attn.q_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.12.self_attn.k_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.12.self_attn.v_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.12.self_attn.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.12.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.12.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.12.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.13.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.13.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.13.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.13.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.13.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.14.self_attn.q_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.14.self_attn.k_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.14.self_attn.v_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 512,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.14.self_attn.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.14.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.14.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.14.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.15.conv.in_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 3072,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.15.conv.out_proj": {
"kind": "linear",
"in_features": 1024,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.15.feed_forward.w1": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.15.feed_forward.w3": {
"kind": "linear",
"in_features": 1024,
"out_features": 4608,
"group_size": 32,
"bias": false,
"recovery_rank": 128
},
"layers.15.feed_forward.w2": {
"kind": "linear",
"in_features": 4608,
"out_features": 1024,
"group_size": 32,
"bias": false,
"recovery_rank": 128
}
},
"base_revision": "b886781f7c6f10ca9b7096e21b83e30a073c2f39"
}