latest
2.5GB
Q4_K_M.gguf https://huggingface.co/openbmb/MiniCPM3-4B
674 Pulls Updated 2 weeks ago
dc14ec45c23d · 2.5GB
-
general.architectureminicpm3
-
general.file_typeQ4_K_M
-
minicpm3.attention.head_count40
-
minicpm3.attention.head_count_kv40
-
minicpm3.attention.key_length96
-
minicpm3.attention.kv_lora_rank256
-
minicpm3.attention.layer_norm_rms_epsilon1e-05
-
minicpm3.attention.q_lora_rank768
-
minicpm3.block_count62
-
minicpm3.context_length32768
-
minicpm3.embedding_length2560
-
minicpm3.feed_forward_length6400
-
minicpm3.rope.dimension_count32
-
minicpm3.vocab_size73448
-
tokenizer.ggml.add_bos_tokentrue
-
tokenizer.ggml.add_eos_tokenfalse
-
tokenizer.ggml.bos_token_id1
-
tokenizer.ggml.eos_token_id73440
-
tokenizer.ggml.modelllama
-
tokenizer.ggml.predefault
-
tokenizer.ggml.scores[-1000 -1000 -1000 -1000 -1000 ...]
-
tokenizer.ggml.token_type[3 3 1 1 1 ...]
-
tokenizer.ggml.tokens[<unk> <s> </s> <SEP> <CLS> ...]
-
tokenizer.ggml.unknown_token_id0
-
NameTypeShape
-
token_embd.weightQ6_K[2560 73448]
-
blk.0.attn_norm.weightF32[2560]
-
blk.0.attn_q_b.weightQ4_K[768 3840]
-
blk.0.attn_q_a.weightQ4_K[2560 768]
-
blk.0.attn_kv_b.weightQ4_K[256 5120]
-
blk.0.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.0.attn_q_a_norm.weightF32[768]
-
blk.0.attn_kv_a_norm.weightF32[256]
-
blk.0.attn_output.weightQ4_K[2560 2560]
-
blk.0.ffn_norm.weightF32[2560]
-
blk.0.ffn_gate.weightQ4_K[2560 6400]
-
blk.0.ffn_up.weightQ4_K[2560 6400]
-
blk.0.ffn_down.weightQ6_K[6400 2560]
-
blk.1.attn_norm.weightF32[2560]
-
blk.1.attn_q_b.weightQ4_K[768 3840]
-
blk.1.attn_q_a.weightQ4_K[2560 768]
-
blk.1.attn_kv_b.weightQ4_K[256 5120]
-
blk.1.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.1.attn_q_a_norm.weightF32[768]
-
blk.1.attn_kv_a_norm.weightF32[256]
-
blk.1.attn_output.weightQ4_K[2560 2560]
-
blk.1.ffn_norm.weightF32[2560]
-
blk.1.ffn_gate.weightQ4_K[2560 6400]
-
blk.1.ffn_up.weightQ4_K[2560 6400]
-
blk.1.ffn_down.weightQ6_K[6400 2560]
-
blk.2.attn_norm.weightF32[2560]
-
blk.2.attn_q_b.weightQ4_K[768 3840]
-
blk.2.attn_q_a.weightQ4_K[2560 768]
-
blk.2.attn_kv_b.weightQ4_K[256 5120]
-
blk.2.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.2.attn_q_a_norm.weightF32[768]
-
blk.2.attn_kv_a_norm.weightF32[256]
-
blk.2.attn_output.weightQ4_K[2560 2560]
-
blk.2.ffn_norm.weightF32[2560]
-
blk.2.ffn_gate.weightQ4_K[2560 6400]
-
blk.2.ffn_up.weightQ4_K[2560 6400]
-
blk.2.ffn_down.weightQ6_K[6400 2560]
-
blk.3.attn_norm.weightF32[2560]
-
blk.3.attn_q_b.weightQ4_K[768 3840]
-
blk.3.attn_q_a.weightQ4_K[2560 768]
-
blk.3.attn_kv_b.weightQ4_K[256 5120]
-
blk.3.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.3.attn_q_a_norm.weightF32[768]
-
blk.3.attn_kv_a_norm.weightF32[256]
-
blk.3.attn_output.weightQ4_K[2560 2560]
-
blk.3.ffn_norm.weightF32[2560]
-
blk.3.ffn_gate.weightQ4_K[2560 6400]
-
blk.3.ffn_up.weightQ4_K[2560 6400]
-
blk.3.ffn_down.weightQ6_K[6400 2560]
-
blk.4.attn_norm.weightF32[2560]
-
blk.4.attn_q_b.weightQ4_K[768 3840]
-
blk.4.attn_q_a.weightQ4_K[2560 768]
-
blk.4.attn_kv_b.weightQ4_K[256 5120]
-
blk.4.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.4.attn_q_a_norm.weightF32[768]
-
blk.4.attn_kv_a_norm.weightF32[256]
-
blk.4.attn_output.weightQ4_K[2560 2560]
-
blk.4.ffn_norm.weightF32[2560]
-
blk.4.ffn_gate.weightQ4_K[2560 6400]
-
blk.4.ffn_up.weightQ4_K[2560 6400]
-
blk.4.ffn_down.weightQ6_K[6400 2560]
-
blk.5.attn_norm.weightF32[2560]
-
blk.5.attn_q_b.weightQ4_K[768 3840]
-
blk.5.attn_q_a.weightQ4_K[2560 768]
-
blk.5.attn_kv_b.weightQ4_K[256 5120]
-
blk.5.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.5.attn_q_a_norm.weightF32[768]
-
blk.5.attn_kv_a_norm.weightF32[256]
-
blk.5.attn_output.weightQ4_K[2560 2560]
-
blk.5.ffn_norm.weightF32[2560]
-
blk.5.ffn_gate.weightQ4_K[2560 6400]
-
blk.5.ffn_up.weightQ4_K[2560 6400]
-
blk.5.ffn_down.weightQ6_K[6400 2560]
-
blk.6.attn_norm.weightF32[2560]
-
blk.6.attn_q_b.weightQ4_K[768 3840]
-
blk.6.attn_q_a.weightQ4_K[2560 768]
-
blk.6.attn_kv_b.weightQ4_K[256 5120]
-
blk.6.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.6.attn_q_a_norm.weightF32[768]
-
blk.6.attn_kv_a_norm.weightF32[256]
-
blk.6.attn_output.weightQ4_K[2560 2560]
-
blk.6.ffn_norm.weightF32[2560]
-
blk.6.ffn_gate.weightQ4_K[2560 6400]
-
blk.6.ffn_up.weightQ4_K[2560 6400]
-
blk.6.ffn_down.weightQ6_K[6400 2560]
-
blk.7.attn_norm.weightF32[2560]
-
blk.7.attn_q_b.weightQ4_K[768 3840]
-
blk.7.attn_q_a.weightQ4_K[2560 768]
-
blk.7.attn_kv_b.weightQ4_K[256 5120]
-
blk.7.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.7.attn_q_a_norm.weightF32[768]
-
blk.7.attn_kv_a_norm.weightF32[256]
-
blk.7.attn_output.weightQ4_K[2560 2560]
-
blk.7.ffn_norm.weightF32[2560]
-
blk.7.ffn_gate.weightQ4_K[2560 6400]
-
blk.7.ffn_up.weightQ4_K[2560 6400]
-
blk.7.ffn_down.weightQ4_K[6400 2560]
-
blk.8.attn_norm.weightF32[2560]
-
blk.8.attn_q_b.weightQ4_K[768 3840]
-
blk.8.attn_q_a.weightQ4_K[2560 768]
-
blk.8.attn_kv_b.weightQ4_K[256 5120]
-
blk.8.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.8.attn_q_a_norm.weightF32[768]
-
blk.8.attn_kv_a_norm.weightF32[256]
-
blk.8.attn_output.weightQ4_K[2560 2560]
-
blk.8.ffn_norm.weightF32[2560]
-
blk.8.ffn_gate.weightQ4_K[2560 6400]
-
blk.8.ffn_up.weightQ4_K[2560 6400]
-
blk.8.ffn_down.weightQ4_K[6400 2560]
-
blk.9.attn_norm.weightF32[2560]
-
blk.9.attn_q_b.weightQ4_K[768 3840]
-
blk.9.attn_q_a.weightQ4_K[2560 768]
-
blk.9.attn_kv_b.weightQ4_K[256 5120]
-
blk.9.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.9.attn_q_a_norm.weightF32[768]
-
blk.9.attn_kv_a_norm.weightF32[256]
-
blk.9.attn_output.weightQ4_K[2560 2560]
-
blk.9.ffn_norm.weightF32[2560]
-
blk.9.ffn_gate.weightQ4_K[2560 6400]
-
blk.9.ffn_up.weightQ4_K[2560 6400]
-
blk.9.ffn_down.weightQ6_K[6400 2560]
-
blk.10.attn_norm.weightF32[2560]
-
blk.10.attn_q_b.weightQ4_K[768 3840]
-
blk.10.attn_q_a.weightQ4_K[2560 768]
-
blk.10.attn_kv_b.weightQ4_K[256 5120]
-
blk.10.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.10.attn_q_a_norm.weightF32[768]
-
blk.10.attn_kv_a_norm.weightF32[256]
-
blk.10.attn_output.weightQ4_K[2560 2560]
-
blk.10.ffn_norm.weightF32[2560]
-
blk.10.ffn_gate.weightQ4_K[2560 6400]
-
blk.10.ffn_up.weightQ4_K[2560 6400]
-
blk.10.ffn_down.weightQ4_K[6400 2560]
-
blk.11.attn_norm.weightF32[2560]
-
blk.11.attn_q_b.weightQ4_K[768 3840]
-
blk.11.attn_q_a.weightQ4_K[2560 768]
-
blk.11.attn_kv_b.weightQ4_K[256 5120]
-
blk.11.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.11.attn_q_a_norm.weightF32[768]
-
blk.11.attn_kv_a_norm.weightF32[256]
-
blk.11.attn_output.weightQ4_K[2560 2560]
-
blk.11.ffn_norm.weightF32[2560]
-
blk.11.ffn_gate.weightQ4_K[2560 6400]
-
blk.11.ffn_up.weightQ4_K[2560 6400]
-
blk.11.ffn_down.weightQ4_K[6400 2560]
-
blk.12.attn_norm.weightF32[2560]
-
blk.12.attn_q_b.weightQ4_K[768 3840]
-
blk.12.attn_q_a.weightQ4_K[2560 768]
-
blk.12.attn_kv_b.weightQ4_K[256 5120]
-
blk.12.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.12.attn_q_a_norm.weightF32[768]
-
blk.12.attn_kv_a_norm.weightF32[256]
-
blk.12.attn_output.weightQ4_K[2560 2560]
-
blk.12.ffn_norm.weightF32[2560]
-
blk.12.ffn_gate.weightQ4_K[2560 6400]
-
blk.12.ffn_up.weightQ4_K[2560 6400]
-
blk.12.ffn_down.weightQ6_K[6400 2560]
-
blk.13.attn_norm.weightF32[2560]
-
blk.13.attn_q_b.weightQ4_K[768 3840]
-
blk.13.attn_q_a.weightQ4_K[2560 768]
-
blk.13.attn_kv_b.weightQ4_K[256 5120]
-
blk.13.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.13.attn_q_a_norm.weightF32[768]
-
blk.13.attn_kv_a_norm.weightF32[256]
-
blk.13.attn_output.weightQ4_K[2560 2560]
-
blk.13.ffn_norm.weightF32[2560]
-
blk.13.ffn_gate.weightQ4_K[2560 6400]
-
blk.13.ffn_up.weightQ4_K[2560 6400]
-
blk.13.ffn_down.weightQ4_K[6400 2560]
-
blk.14.attn_norm.weightF32[2560]
-
blk.14.attn_q_b.weightQ4_K[768 3840]
-
blk.14.attn_q_a.weightQ4_K[2560 768]
-
blk.14.attn_kv_b.weightQ4_K[256 5120]
-
blk.14.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.14.attn_q_a_norm.weightF32[768]
-
blk.14.attn_kv_a_norm.weightF32[256]
-
blk.14.attn_output.weightQ4_K[2560 2560]
-
blk.14.ffn_norm.weightF32[2560]
-
blk.14.ffn_gate.weightQ4_K[2560 6400]
-
blk.14.ffn_up.weightQ4_K[2560 6400]
-
blk.14.ffn_down.weightQ4_K[6400 2560]
-
blk.15.attn_norm.weightF32[2560]
-
blk.15.attn_q_b.weightQ4_K[768 3840]
-
blk.15.attn_q_a.weightQ4_K[2560 768]
-
blk.15.attn_kv_b.weightQ4_K[256 5120]
-
blk.15.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.15.attn_q_a_norm.weightF32[768]
-
blk.15.attn_kv_a_norm.weightF32[256]
-
blk.15.attn_output.weightQ4_K[2560 2560]
-
blk.15.ffn_norm.weightF32[2560]
-
blk.15.ffn_gate.weightQ4_K[2560 6400]
-
blk.15.ffn_up.weightQ4_K[2560 6400]
-
blk.15.ffn_down.weightQ6_K[6400 2560]
-
blk.16.attn_norm.weightF32[2560]
-
blk.16.attn_q_b.weightQ4_K[768 3840]
-
blk.16.attn_q_a.weightQ4_K[2560 768]
-
blk.16.attn_kv_b.weightQ4_K[256 5120]
-
blk.16.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.16.attn_q_a_norm.weightF32[768]
-
blk.16.attn_kv_a_norm.weightF32[256]
-
blk.16.attn_output.weightQ4_K[2560 2560]
-
blk.16.ffn_norm.weightF32[2560]
-
blk.16.ffn_gate.weightQ4_K[2560 6400]
-
blk.16.ffn_up.weightQ4_K[2560 6400]
-
blk.16.ffn_down.weightQ4_K[6400 2560]
-
blk.17.attn_norm.weightF32[2560]
-
blk.17.attn_q_b.weightQ4_K[768 3840]
-
blk.17.attn_q_a.weightQ4_K[2560 768]
-
blk.17.attn_kv_b.weightQ4_K[256 5120]
-
blk.17.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.17.attn_q_a_norm.weightF32[768]
-
blk.17.attn_kv_a_norm.weightF32[256]
-
blk.17.attn_output.weightQ4_K[2560 2560]
-
blk.17.ffn_norm.weightF32[2560]
-
blk.17.ffn_gate.weightQ4_K[2560 6400]
-
blk.17.ffn_up.weightQ4_K[2560 6400]
-
blk.17.ffn_down.weightQ4_K[6400 2560]
-
blk.18.attn_norm.weightF32[2560]
-
blk.18.attn_q_b.weightQ4_K[768 3840]
-
blk.18.attn_q_a.weightQ4_K[2560 768]
-
blk.18.attn_kv_b.weightQ4_K[256 5120]
-
blk.18.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.18.attn_q_a_norm.weightF32[768]
-
blk.18.attn_kv_a_norm.weightF32[256]
-
blk.18.attn_output.weightQ4_K[2560 2560]
-
blk.18.ffn_norm.weightF32[2560]
-
blk.18.ffn_gate.weightQ4_K[2560 6400]
-
blk.18.ffn_up.weightQ4_K[2560 6400]
-
blk.18.ffn_down.weightQ6_K[6400 2560]
-
blk.19.attn_norm.weightF32[2560]
-
blk.19.attn_q_b.weightQ4_K[768 3840]
-
blk.19.attn_q_a.weightQ4_K[2560 768]
-
blk.19.attn_kv_b.weightQ4_K[256 5120]
-
blk.19.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.19.attn_q_a_norm.weightF32[768]
-
blk.19.attn_kv_a_norm.weightF32[256]
-
blk.19.attn_output.weightQ4_K[2560 2560]
-
blk.19.ffn_norm.weightF32[2560]
-
blk.19.ffn_gate.weightQ4_K[2560 6400]
-
blk.19.ffn_up.weightQ4_K[2560 6400]
-
blk.19.ffn_down.weightQ4_K[6400 2560]
-
blk.20.attn_norm.weightF32[2560]
-
blk.20.attn_q_b.weightQ4_K[768 3840]
-
blk.20.attn_q_a.weightQ4_K[2560 768]
-
blk.20.attn_kv_b.weightQ4_K[256 5120]
-
blk.20.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.20.attn_q_a_norm.weightF32[768]
-
blk.20.attn_kv_a_norm.weightF32[256]
-
blk.20.attn_output.weightQ4_K[2560 2560]
-
blk.20.ffn_norm.weightF32[2560]
-
blk.20.ffn_gate.weightQ4_K[2560 6400]
-
blk.20.ffn_up.weightQ4_K[2560 6400]
-
blk.20.ffn_down.weightQ4_K[6400 2560]
-
blk.21.attn_norm.weightF32[2560]
-
blk.21.attn_q_b.weightQ4_K[768 3840]
-
blk.21.attn_q_a.weightQ4_K[2560 768]
-
blk.21.attn_kv_b.weightQ4_K[256 5120]
-
blk.21.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.21.attn_q_a_norm.weightF32[768]
-
blk.21.attn_kv_a_norm.weightF32[256]
-
blk.21.attn_output.weightQ4_K[2560 2560]
-
blk.21.ffn_norm.weightF32[2560]
-
blk.21.ffn_gate.weightQ4_K[2560 6400]
-
blk.21.ffn_up.weightQ4_K[2560 6400]
-
blk.21.ffn_down.weightQ6_K[6400 2560]
-
blk.22.attn_norm.weightF32[2560]
-
blk.22.attn_q_b.weightQ4_K[768 3840]
-
blk.22.attn_q_a.weightQ4_K[2560 768]
-
blk.22.attn_kv_b.weightQ4_K[256 5120]
-
blk.22.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.22.attn_q_a_norm.weightF32[768]
-
blk.22.attn_kv_a_norm.weightF32[256]
-
blk.22.attn_output.weightQ4_K[2560 2560]
-
blk.22.ffn_norm.weightF32[2560]
-
blk.22.ffn_gate.weightQ4_K[2560 6400]
-
blk.22.ffn_up.weightQ4_K[2560 6400]
-
blk.22.ffn_down.weightQ4_K[6400 2560]
-
blk.23.attn_norm.weightF32[2560]
-
blk.23.attn_q_b.weightQ4_K[768 3840]
-
blk.23.attn_q_a.weightQ4_K[2560 768]
-
blk.23.attn_kv_b.weightQ4_K[256 5120]
-
blk.23.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.23.attn_q_a_norm.weightF32[768]
-
blk.23.attn_kv_a_norm.weightF32[256]
-
blk.23.attn_output.weightQ4_K[2560 2560]
-
blk.23.ffn_norm.weightF32[2560]
-
blk.23.ffn_gate.weightQ4_K[2560 6400]
-
blk.23.ffn_up.weightQ4_K[2560 6400]
-
blk.23.ffn_down.weightQ4_K[6400 2560]
-
blk.24.attn_norm.weightF32[2560]
-
blk.24.attn_q_b.weightQ4_K[768 3840]
-
blk.24.attn_q_a.weightQ4_K[2560 768]
-
blk.24.attn_kv_b.weightQ4_K[256 5120]
-
blk.24.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.24.attn_q_a_norm.weightF32[768]
-
blk.24.attn_kv_a_norm.weightF32[256]
-
blk.24.attn_output.weightQ4_K[2560 2560]
-
blk.24.ffn_norm.weightF32[2560]
-
blk.24.ffn_gate.weightQ4_K[2560 6400]
-
blk.24.ffn_up.weightQ4_K[2560 6400]
-
blk.24.ffn_down.weightQ6_K[6400 2560]
-
blk.25.attn_norm.weightF32[2560]
-
blk.25.attn_q_b.weightQ4_K[768 3840]
-
blk.25.attn_q_a.weightQ4_K[2560 768]
-
blk.25.attn_kv_b.weightQ4_K[256 5120]
-
blk.25.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.25.attn_q_a_norm.weightF32[768]
-
blk.25.attn_kv_a_norm.weightF32[256]
-
blk.25.attn_output.weightQ4_K[2560 2560]
-
blk.25.ffn_norm.weightF32[2560]
-
blk.25.ffn_gate.weightQ4_K[2560 6400]
-
blk.25.ffn_up.weightQ4_K[2560 6400]
-
blk.25.ffn_down.weightQ4_K[6400 2560]
-
blk.26.attn_norm.weightF32[2560]
-
blk.26.attn_q_b.weightQ4_K[768 3840]
-
blk.26.attn_q_a.weightQ4_K[2560 768]
-
blk.26.attn_kv_b.weightQ4_K[256 5120]
-
blk.26.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.26.attn_q_a_norm.weightF32[768]
-
blk.26.attn_kv_a_norm.weightF32[256]
-
blk.26.attn_output.weightQ4_K[2560 2560]
-
blk.26.ffn_norm.weightF32[2560]
-
blk.26.ffn_gate.weightQ4_K[2560 6400]
-
blk.26.ffn_up.weightQ4_K[2560 6400]
-
blk.26.ffn_down.weightQ4_K[6400 2560]
-
blk.27.attn_norm.weightF32[2560]
-
blk.27.attn_q_b.weightQ4_K[768 3840]
-
blk.27.attn_q_a.weightQ4_K[2560 768]
-
blk.27.attn_kv_b.weightQ4_K[256 5120]
-
blk.27.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.27.attn_q_a_norm.weightF32[768]
-
blk.27.attn_kv_a_norm.weightF32[256]
-
blk.27.attn_output.weightQ4_K[2560 2560]
-
blk.27.ffn_norm.weightF32[2560]
-
blk.27.ffn_gate.weightQ4_K[2560 6400]
-
blk.27.ffn_up.weightQ4_K[2560 6400]
-
blk.27.ffn_down.weightQ6_K[6400 2560]
-
blk.28.attn_norm.weightF32[2560]
-
blk.28.attn_q_b.weightQ4_K[768 3840]
-
blk.28.attn_q_a.weightQ4_K[2560 768]
-
blk.28.attn_kv_b.weightQ4_K[256 5120]
-
blk.28.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.28.attn_q_a_norm.weightF32[768]
-
blk.28.attn_kv_a_norm.weightF32[256]
-
blk.28.attn_output.weightQ4_K[2560 2560]
-
blk.28.ffn_norm.weightF32[2560]
-
blk.28.ffn_gate.weightQ4_K[2560 6400]
-
blk.28.ffn_up.weightQ4_K[2560 6400]
-
blk.28.ffn_down.weightQ4_K[6400 2560]
-
blk.29.attn_norm.weightF32[2560]
-
blk.29.attn_q_b.weightQ4_K[768 3840]
-
blk.29.attn_q_a.weightQ4_K[2560 768]
-
blk.29.attn_kv_b.weightQ4_K[256 5120]
-
blk.29.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.29.attn_q_a_norm.weightF32[768]
-
blk.29.attn_kv_a_norm.weightF32[256]
-
blk.29.attn_output.weightQ4_K[2560 2560]
-
blk.29.ffn_norm.weightF32[2560]
-
blk.29.ffn_gate.weightQ4_K[2560 6400]
-
blk.29.ffn_up.weightQ4_K[2560 6400]
-
blk.29.ffn_down.weightQ4_K[6400 2560]
-
blk.30.attn_norm.weightF32[2560]
-
blk.30.attn_q_b.weightQ4_K[768 3840]
-
blk.30.attn_q_a.weightQ4_K[2560 768]
-
blk.30.attn_kv_b.weightQ4_K[256 5120]
-
blk.30.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.30.attn_q_a_norm.weightF32[768]
-
blk.30.attn_kv_a_norm.weightF32[256]
-
blk.30.attn_output.weightQ4_K[2560 2560]
-
blk.30.ffn_norm.weightF32[2560]
-
blk.30.ffn_gate.weightQ4_K[2560 6400]
-
blk.30.ffn_up.weightQ4_K[2560 6400]
-
blk.30.ffn_down.weightQ6_K[6400 2560]
-
blk.31.attn_norm.weightF32[2560]
-
blk.31.attn_q_b.weightQ4_K[768 3840]
-
blk.31.attn_q_a.weightQ4_K[2560 768]
-
blk.31.attn_kv_b.weightQ4_K[256 5120]
-
blk.31.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.31.attn_q_a_norm.weightF32[768]
-
blk.31.attn_kv_a_norm.weightF32[256]
-
blk.31.attn_output.weightQ4_K[2560 2560]
-
blk.31.ffn_norm.weightF32[2560]
-
blk.31.ffn_gate.weightQ4_K[2560 6400]
-
blk.31.ffn_up.weightQ4_K[2560 6400]
-
blk.31.ffn_down.weightQ4_K[6400 2560]
-
blk.32.attn_norm.weightF32[2560]
-
blk.32.attn_q_b.weightQ4_K[768 3840]
-
blk.32.attn_q_a.weightQ4_K[2560 768]
-
blk.32.attn_kv_b.weightQ4_K[256 5120]
-
blk.32.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.32.attn_q_a_norm.weightF32[768]
-
blk.32.attn_kv_a_norm.weightF32[256]
-
blk.32.attn_output.weightQ4_K[2560 2560]
-
blk.32.ffn_norm.weightF32[2560]
-
blk.32.ffn_gate.weightQ4_K[2560 6400]
-
blk.32.ffn_up.weightQ4_K[2560 6400]
-
blk.32.ffn_down.weightQ4_K[6400 2560]
-
blk.33.attn_norm.weightF32[2560]
-
blk.33.attn_q_b.weightQ4_K[768 3840]
-
blk.33.attn_q_a.weightQ4_K[2560 768]
-
blk.33.attn_kv_b.weightQ4_K[256 5120]
-
blk.33.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.33.attn_q_a_norm.weightF32[768]
-
blk.33.attn_kv_a_norm.weightF32[256]
-
blk.33.attn_output.weightQ4_K[2560 2560]
-
blk.33.ffn_norm.weightF32[2560]
-
blk.33.ffn_gate.weightQ4_K[2560 6400]
-
blk.33.ffn_up.weightQ4_K[2560 6400]
-
blk.33.ffn_down.weightQ6_K[6400 2560]
-
blk.34.attn_norm.weightF32[2560]
-
blk.34.attn_q_b.weightQ4_K[768 3840]
-
blk.34.attn_q_a.weightQ4_K[2560 768]
-
blk.34.attn_kv_b.weightQ4_K[256 5120]
-
blk.34.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.34.attn_q_a_norm.weightF32[768]
-
blk.34.attn_kv_a_norm.weightF32[256]
-
blk.34.attn_output.weightQ4_K[2560 2560]
-
blk.34.ffn_norm.weightF32[2560]
-
blk.34.ffn_gate.weightQ4_K[2560 6400]
-
blk.34.ffn_up.weightQ4_K[2560 6400]
-
blk.34.ffn_down.weightQ4_K[6400 2560]
-
blk.35.attn_norm.weightF32[2560]
-
blk.35.attn_q_b.weightQ4_K[768 3840]
-
blk.35.attn_q_a.weightQ4_K[2560 768]
-
blk.35.attn_kv_b.weightQ4_K[256 5120]
-
blk.35.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.35.attn_q_a_norm.weightF32[768]
-
blk.35.attn_kv_a_norm.weightF32[256]
-
blk.35.attn_output.weightQ4_K[2560 2560]
-
blk.35.ffn_norm.weightF32[2560]
-
blk.35.ffn_gate.weightQ4_K[2560 6400]
-
blk.35.ffn_up.weightQ4_K[2560 6400]
-
blk.35.ffn_down.weightQ4_K[6400 2560]
-
blk.36.attn_norm.weightF32[2560]
-
blk.36.attn_q_b.weightQ4_K[768 3840]
-
blk.36.attn_q_a.weightQ4_K[2560 768]
-
blk.36.attn_kv_b.weightQ4_K[256 5120]
-
blk.36.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.36.attn_q_a_norm.weightF32[768]
-
blk.36.attn_kv_a_norm.weightF32[256]
-
blk.36.attn_output.weightQ4_K[2560 2560]
-
blk.36.ffn_norm.weightF32[2560]
-
blk.36.ffn_gate.weightQ4_K[2560 6400]
-
blk.36.ffn_up.weightQ4_K[2560 6400]
-
blk.36.ffn_down.weightQ6_K[6400 2560]
-
blk.37.attn_norm.weightF32[2560]
-
blk.37.attn_q_b.weightQ4_K[768 3840]
-
blk.37.attn_q_a.weightQ4_K[2560 768]
-
blk.37.attn_kv_b.weightQ4_K[256 5120]
-
blk.37.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.37.attn_q_a_norm.weightF32[768]
-
blk.37.attn_kv_a_norm.weightF32[256]
-
blk.37.attn_output.weightQ4_K[2560 2560]
-
blk.37.ffn_norm.weightF32[2560]
-
blk.37.ffn_gate.weightQ4_K[2560 6400]
-
blk.37.ffn_up.weightQ4_K[2560 6400]
-
blk.37.ffn_down.weightQ4_K[6400 2560]
-
blk.38.attn_norm.weightF32[2560]
-
blk.38.attn_q_b.weightQ4_K[768 3840]
-
blk.38.attn_q_a.weightQ4_K[2560 768]
-
blk.38.attn_kv_b.weightQ4_K[256 5120]
-
blk.38.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.38.attn_q_a_norm.weightF32[768]
-
blk.38.attn_kv_a_norm.weightF32[256]
-
blk.38.attn_output.weightQ4_K[2560 2560]
-
blk.38.ffn_norm.weightF32[2560]
-
blk.38.ffn_gate.weightQ4_K[2560 6400]
-
blk.38.ffn_up.weightQ4_K[2560 6400]
-
blk.38.ffn_down.weightQ4_K[6400 2560]
-
blk.39.attn_norm.weightF32[2560]
-
blk.39.attn_q_b.weightQ4_K[768 3840]
-
blk.39.attn_q_a.weightQ4_K[2560 768]
-
blk.39.attn_kv_b.weightQ4_K[256 5120]
-
blk.39.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.39.attn_q_a_norm.weightF32[768]
-
blk.39.attn_kv_a_norm.weightF32[256]
-
blk.39.attn_output.weightQ4_K[2560 2560]
-
blk.39.ffn_norm.weightF32[2560]
-
blk.39.ffn_gate.weightQ4_K[2560 6400]
-
blk.39.ffn_up.weightQ4_K[2560 6400]
-
blk.39.ffn_down.weightQ6_K[6400 2560]
-
blk.40.attn_norm.weightF32[2560]
-
blk.40.attn_q_b.weightQ4_K[768 3840]
-
blk.40.attn_q_a.weightQ4_K[2560 768]
-
blk.40.attn_kv_b.weightQ4_K[256 5120]
-
blk.40.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.40.attn_q_a_norm.weightF32[768]
-
blk.40.attn_kv_a_norm.weightF32[256]
-
blk.40.attn_output.weightQ4_K[2560 2560]
-
blk.40.ffn_norm.weightF32[2560]
-
blk.40.ffn_gate.weightQ4_K[2560 6400]
-
blk.40.ffn_up.weightQ4_K[2560 6400]
-
blk.40.ffn_down.weightQ4_K[6400 2560]
-
blk.41.attn_norm.weightF32[2560]
-
blk.41.attn_q_b.weightQ4_K[768 3840]
-
blk.41.attn_q_a.weightQ4_K[2560 768]
-
blk.41.attn_kv_b.weightQ4_K[256 5120]
-
blk.41.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.41.attn_q_a_norm.weightF32[768]
-
blk.41.attn_kv_a_norm.weightF32[256]
-
blk.41.attn_output.weightQ4_K[2560 2560]
-
blk.41.ffn_norm.weightF32[2560]
-
blk.41.ffn_gate.weightQ4_K[2560 6400]
-
blk.41.ffn_up.weightQ4_K[2560 6400]
-
blk.41.ffn_down.weightQ4_K[6400 2560]
-
blk.42.attn_norm.weightF32[2560]
-
blk.42.attn_q_b.weightQ4_K[768 3840]
-
blk.42.attn_q_a.weightQ4_K[2560 768]
-
blk.42.attn_kv_b.weightQ4_K[256 5120]
-
blk.42.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.42.attn_q_a_norm.weightF32[768]
-
blk.42.attn_kv_a_norm.weightF32[256]
-
blk.42.attn_output.weightQ4_K[2560 2560]
-
blk.42.ffn_norm.weightF32[2560]
-
blk.42.ffn_gate.weightQ4_K[2560 6400]
-
blk.42.ffn_up.weightQ4_K[2560 6400]
-
blk.42.ffn_down.weightQ6_K[6400 2560]
-
blk.43.attn_norm.weightF32[2560]
-
blk.43.attn_q_b.weightQ4_K[768 3840]
-
blk.43.attn_q_a.weightQ4_K[2560 768]
-
blk.43.attn_kv_b.weightQ4_K[256 5120]
-
blk.43.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.43.attn_q_a_norm.weightF32[768]
-
blk.43.attn_kv_a_norm.weightF32[256]
-
blk.43.attn_output.weightQ4_K[2560 2560]
-
blk.43.ffn_norm.weightF32[2560]
-
blk.43.ffn_gate.weightQ4_K[2560 6400]
-
blk.43.ffn_up.weightQ4_K[2560 6400]
-
blk.43.ffn_down.weightQ4_K[6400 2560]
-
blk.44.attn_norm.weightF32[2560]
-
blk.44.attn_q_b.weightQ4_K[768 3840]
-
blk.44.attn_q_a.weightQ4_K[2560 768]
-
blk.44.attn_kv_b.weightQ4_K[256 5120]
-
blk.44.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.44.attn_q_a_norm.weightF32[768]
-
blk.44.attn_kv_a_norm.weightF32[256]
-
blk.44.attn_output.weightQ4_K[2560 2560]
-
blk.44.ffn_norm.weightF32[2560]
-
blk.44.ffn_gate.weightQ4_K[2560 6400]
-
blk.44.ffn_up.weightQ4_K[2560 6400]
-
blk.44.ffn_down.weightQ4_K[6400 2560]
-
blk.45.attn_norm.weightF32[2560]
-
blk.45.attn_q_b.weightQ4_K[768 3840]
-
blk.45.attn_q_a.weightQ4_K[2560 768]
-
blk.45.attn_kv_b.weightQ4_K[256 5120]
-
blk.45.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.45.attn_q_a_norm.weightF32[768]
-
blk.45.attn_kv_a_norm.weightF32[256]
-
blk.45.attn_output.weightQ4_K[2560 2560]
-
blk.45.ffn_norm.weightF32[2560]
-
blk.45.ffn_gate.weightQ4_K[2560 6400]
-
blk.45.ffn_up.weightQ4_K[2560 6400]
-
blk.45.ffn_down.weightQ6_K[6400 2560]
-
blk.46.attn_norm.weightF32[2560]
-
blk.46.attn_q_b.weightQ4_K[768 3840]
-
blk.46.attn_q_a.weightQ4_K[2560 768]
-
blk.46.attn_kv_b.weightQ4_K[256 5120]
-
blk.46.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.46.attn_q_a_norm.weightF32[768]
-
blk.46.attn_kv_a_norm.weightF32[256]
-
blk.46.attn_output.weightQ4_K[2560 2560]
-
blk.46.ffn_norm.weightF32[2560]
-
blk.46.ffn_gate.weightQ4_K[2560 6400]
-
blk.46.ffn_up.weightQ4_K[2560 6400]
-
blk.46.ffn_down.weightQ4_K[6400 2560]
-
blk.47.attn_norm.weightF32[2560]
-
blk.47.attn_q_b.weightQ4_K[768 3840]
-
blk.47.attn_q_a.weightQ4_K[2560 768]
-
blk.47.attn_kv_b.weightQ4_K[256 5120]
-
blk.47.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.47.attn_q_a_norm.weightF32[768]
-
blk.47.attn_kv_a_norm.weightF32[256]
-
blk.47.attn_output.weightQ4_K[2560 2560]
-
blk.47.ffn_norm.weightF32[2560]
-
blk.47.ffn_gate.weightQ4_K[2560 6400]
-
blk.47.ffn_up.weightQ4_K[2560 6400]
-
blk.47.ffn_down.weightQ4_K[6400 2560]
-
blk.48.attn_norm.weightF32[2560]
-
blk.48.attn_q_b.weightQ4_K[768 3840]
-
blk.48.attn_q_a.weightQ4_K[2560 768]
-
blk.48.attn_kv_b.weightQ4_K[256 5120]
-
blk.48.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.48.attn_q_a_norm.weightF32[768]
-
blk.48.attn_kv_a_norm.weightF32[256]
-
blk.48.attn_output.weightQ4_K[2560 2560]
-
blk.48.ffn_norm.weightF32[2560]
-
blk.48.ffn_gate.weightQ4_K[2560 6400]
-
blk.48.ffn_up.weightQ4_K[2560 6400]
-
blk.48.ffn_down.weightQ6_K[6400 2560]
-
blk.49.attn_norm.weightF32[2560]
-
blk.49.attn_q_b.weightQ4_K[768 3840]
-
blk.49.attn_q_a.weightQ4_K[2560 768]
-
blk.49.attn_kv_b.weightQ4_K[256 5120]
-
blk.49.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.49.attn_q_a_norm.weightF32[768]
-
blk.49.attn_kv_a_norm.weightF32[256]
-
blk.49.attn_output.weightQ4_K[2560 2560]
-
blk.49.ffn_norm.weightF32[2560]
-
blk.49.ffn_gate.weightQ4_K[2560 6400]
-
blk.49.ffn_up.weightQ4_K[2560 6400]
-
blk.49.ffn_down.weightQ4_K[6400 2560]
-
blk.50.attn_norm.weightF32[2560]
-
blk.50.attn_q_b.weightQ4_K[768 3840]
-
blk.50.attn_q_a.weightQ4_K[2560 768]
-
blk.50.attn_kv_b.weightQ4_K[256 5120]
-
blk.50.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.50.attn_q_a_norm.weightF32[768]
-
blk.50.attn_kv_a_norm.weightF32[256]
-
blk.50.attn_output.weightQ4_K[2560 2560]
-
blk.50.ffn_norm.weightF32[2560]
-
blk.50.ffn_gate.weightQ4_K[2560 6400]
-
blk.50.ffn_up.weightQ4_K[2560 6400]
-
blk.50.ffn_down.weightQ4_K[6400 2560]
-
blk.51.attn_norm.weightF32[2560]
-
blk.51.attn_q_b.weightQ4_K[768 3840]
-
blk.51.attn_q_a.weightQ4_K[2560 768]
-
blk.51.attn_kv_b.weightQ4_K[256 5120]
-
blk.51.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.51.attn_q_a_norm.weightF32[768]
-
blk.51.attn_kv_a_norm.weightF32[256]
-
blk.51.attn_output.weightQ4_K[2560 2560]
-
blk.51.ffn_norm.weightF32[2560]
-
blk.51.ffn_gate.weightQ4_K[2560 6400]
-
blk.51.ffn_up.weightQ4_K[2560 6400]
-
blk.51.ffn_down.weightQ6_K[6400 2560]
-
blk.52.attn_norm.weightF32[2560]
-
blk.52.attn_q_b.weightQ4_K[768 3840]
-
blk.52.attn_q_a.weightQ4_K[2560 768]
-
blk.52.attn_kv_b.weightQ4_K[256 5120]
-
blk.52.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.52.attn_q_a_norm.weightF32[768]
-
blk.52.attn_kv_a_norm.weightF32[256]
-
blk.52.attn_output.weightQ4_K[2560 2560]
-
blk.52.ffn_norm.weightF32[2560]
-
blk.52.ffn_gate.weightQ4_K[2560 6400]
-
blk.52.ffn_up.weightQ4_K[2560 6400]
-
blk.52.ffn_down.weightQ4_K[6400 2560]
-
blk.53.attn_norm.weightF32[2560]
-
blk.53.attn_q_b.weightQ4_K[768 3840]
-
blk.53.attn_q_a.weightQ4_K[2560 768]
-
blk.53.attn_kv_b.weightQ4_K[256 5120]
-
blk.53.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.53.attn_q_a_norm.weightF32[768]
-
blk.53.attn_kv_a_norm.weightF32[256]
-
blk.53.attn_output.weightQ4_K[2560 2560]
-
blk.53.ffn_norm.weightF32[2560]
-
blk.53.ffn_gate.weightQ4_K[2560 6400]
-
blk.53.ffn_up.weightQ4_K[2560 6400]
-
blk.53.ffn_down.weightQ4_K[6400 2560]
-
blk.54.attn_norm.weightF32[2560]
-
blk.54.attn_q_b.weightQ4_K[768 3840]
-
blk.54.attn_q_a.weightQ4_K[2560 768]
-
blk.54.attn_kv_b.weightQ4_K[256 5120]
-
blk.54.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.54.attn_q_a_norm.weightF32[768]
-
blk.54.attn_kv_a_norm.weightF32[256]
-
blk.54.attn_output.weightQ4_K[2560 2560]
-
blk.54.ffn_norm.weightF32[2560]
-
blk.54.ffn_gate.weightQ4_K[2560 6400]
-
blk.54.ffn_up.weightQ4_K[2560 6400]
-
blk.54.ffn_down.weightQ6_K[6400 2560]
-
blk.55.attn_norm.weightF32[2560]
-
blk.55.attn_q_b.weightQ4_K[768 3840]
-
blk.55.attn_q_a.weightQ4_K[2560 768]
-
blk.55.attn_kv_b.weightQ4_K[256 5120]
-
blk.55.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.55.attn_q_a_norm.weightF32[768]
-
blk.55.attn_kv_a_norm.weightF32[256]
-
blk.55.attn_output.weightQ4_K[2560 2560]
-
blk.55.ffn_norm.weightF32[2560]
-
blk.55.ffn_gate.weightQ4_K[2560 6400]
-
blk.55.ffn_up.weightQ4_K[2560 6400]
-
blk.55.ffn_down.weightQ6_K[6400 2560]
-
blk.56.attn_norm.weightF32[2560]
-
blk.56.attn_q_b.weightQ4_K[768 3840]
-
blk.56.attn_q_a.weightQ4_K[2560 768]
-
blk.56.attn_kv_b.weightQ4_K[256 5120]
-
blk.56.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.56.attn_q_a_norm.weightF32[768]
-
blk.56.attn_kv_a_norm.weightF32[256]
-
blk.56.attn_output.weightQ4_K[2560 2560]
-
blk.56.ffn_norm.weightF32[2560]
-
blk.56.ffn_gate.weightQ4_K[2560 6400]
-
blk.56.ffn_up.weightQ4_K[2560 6400]
-
blk.56.ffn_down.weightQ6_K[6400 2560]
-
blk.57.attn_norm.weightF32[2560]
-
blk.57.attn_q_b.weightQ4_K[768 3840]
-
blk.57.attn_q_a.weightQ4_K[2560 768]
-
blk.57.attn_kv_b.weightQ4_K[256 5120]
-
blk.57.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.57.attn_q_a_norm.weightF32[768]
-
blk.57.attn_kv_a_norm.weightF32[256]
-
blk.57.attn_output.weightQ4_K[2560 2560]
-
blk.57.ffn_norm.weightF32[2560]
-
blk.57.ffn_gate.weightQ4_K[2560 6400]
-
blk.57.ffn_up.weightQ4_K[2560 6400]
-
blk.57.ffn_down.weightQ6_K[6400 2560]
-
blk.58.attn_norm.weightF32[2560]
-
blk.58.attn_q_b.weightQ4_K[768 3840]
-
blk.58.attn_q_a.weightQ4_K[2560 768]
-
blk.58.attn_kv_b.weightQ4_K[256 5120]
-
blk.58.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.58.attn_q_a_norm.weightF32[768]
-
blk.58.attn_kv_a_norm.weightF32[256]
-
blk.58.attn_output.weightQ4_K[2560 2560]
-
blk.58.ffn_norm.weightF32[2560]
-
blk.58.ffn_gate.weightQ4_K[2560 6400]
-
blk.58.ffn_up.weightQ4_K[2560 6400]
-
blk.58.ffn_down.weightQ6_K[6400 2560]
-
blk.59.attn_norm.weightF32[2560]
-
blk.59.attn_q_b.weightQ4_K[768 3840]
-
blk.59.attn_q_a.weightQ4_K[2560 768]
-
blk.59.attn_kv_b.weightQ4_K[256 5120]
-
blk.59.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.59.attn_q_a_norm.weightF32[768]
-
blk.59.attn_kv_a_norm.weightF32[256]
-
blk.59.attn_output.weightQ4_K[2560 2560]
-
blk.59.ffn_norm.weightF32[2560]
-
blk.59.ffn_gate.weightQ4_K[2560 6400]
-
blk.59.ffn_up.weightQ4_K[2560 6400]
-
blk.59.ffn_down.weightQ6_K[6400 2560]
-
blk.60.attn_norm.weightF32[2560]
-
blk.60.attn_q_b.weightQ4_K[768 3840]
-
blk.60.attn_q_a.weightQ4_K[2560 768]
-
blk.60.attn_kv_b.weightQ4_K[256 5120]
-
blk.60.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.60.attn_q_a_norm.weightF32[768]
-
blk.60.attn_kv_a_norm.weightF32[256]
-
blk.60.attn_output.weightQ4_K[2560 2560]
-
blk.60.ffn_norm.weightF32[2560]
-
blk.60.ffn_gate.weightQ4_K[2560 6400]
-
blk.60.ffn_up.weightQ4_K[2560 6400]
-
blk.60.ffn_down.weightQ6_K[6400 2560]
-
blk.61.attn_norm.weightF32[2560]
-
blk.61.attn_q_b.weightQ4_K[768 3840]
-
blk.61.attn_q_a.weightQ4_K[2560 768]
-
blk.61.attn_kv_b.weightQ4_K[256 5120]
-
blk.61.attn_kv_a_mqa.weightQ4_K[2560 288]
-
blk.61.attn_q_a_norm.weightF32[768]
-
blk.61.attn_kv_a_norm.weightF32[256]
-
blk.61.attn_output.weightQ4_K[2560 2560]
-
blk.61.ffn_norm.weightF32[2560]
-
blk.61.ffn_gate.weightQ4_K[2560 6400]
-
blk.61.ffn_up.weightQ4_K[2560 6400]
-
blk.61.ffn_down.weightQ6_K[6400 2560]
-
output_norm.weightF32[2560]
-
rope_factors_long.weightF32[16]
-
rope_factors_short.weightF32[16]
Metadata
Tensors
blk.0
blk.1
blk.2
blk.3
blk.4
blk.5
blk.6
blk.7
blk.8
blk.9
blk.10
blk.11
blk.12
blk.13
blk.14
blk.15
blk.16
blk.17
blk.18
blk.19
blk.20
blk.21
blk.22
blk.23
blk.24
blk.25
blk.26
blk.27
blk.28
blk.29
blk.30
blk.31
blk.32
blk.33
blk.34
blk.35
blk.36
blk.37
blk.38
blk.39
blk.40
blk.41
blk.42
blk.43
blk.44
blk.45
blk.46
blk.47
blk.48
blk.49
blk.50
blk.51
blk.52
blk.53
blk.54
blk.55
blk.56
blk.57
blk.58
blk.59
blk.60
blk.61