tensor-allocation/DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS.rco-allocation.txt
34.0 KB · 1062 lines · plaintext Raw
1 # RCO per-tensor GGUF allocation
2 # Repository: ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF
3 # GGUF: DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00001-of-00006.gguf (first of 6 shards; this dump covers all of them)
4 # Source: https://huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/main/DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00001-of-00006.gguf
5 # Source: https://huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/main/DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00002-of-00006.gguf
6 # Source: https://huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/main/DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00003-of-00006.gguf
7 # Source: https://huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/main/DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00004-of-00006.gguf
8 # Source: https://huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/main/DeepSeek-V4.1-Flash-1017-GSQ-RCO-IQ2_XXS-00005-of-00006.gguf
9 # Source: https://huggingface.co/ISTA-DASLab/DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF/resolve/main/DeepSeek-V4.1-Flash-Engram-IQ4_NL.gguf
10 # Target bpw over the backbone (all tensors except the Engram table): 2.00 (measured: 2.0038)
11 # The Engram table (per_layer_token_embd, shard 6 = the shared Engram file) is IQ4_NL in every build and not part of the search.
12 # GGUF version: 3
13 # GGUF metadata entries: 77
14 # Tensor count: 1043
15 # Quant-type counts: BF16=67, F32=531, IQ1_M=60, IQ2_XS=18, IQ2_XXS=42, IQ4_NL=1, Q5_K=244, Q8_0=80
16 # Format: <GGUF tensor name>: <GGML quantization/storage type>
17 # Includes fixed non-RCO tensors (e.g. F32/BF16) so the released GGUF is fully reproducible.
18 #
19 output.weight: Q5_K
20 output_norm.weight: F32
21 per_layer_token_embd.weight: IQ4_NL
22 token_embd.weight: Q5_K
23 blk.0.attn_kv.weight: Q8_0
24 blk.0.attn_kv_a_norm.weight: F32
25 blk.0.attn_norm.weight: F32
26 blk.0.attn_output_a.weight: Q5_K
27 blk.0.attn_output_b.weight: Q5_K
28 blk.0.attn_q_a.weight: Q8_0
29 blk.0.attn_q_a_norm.weight: F32
30 blk.0.attn_q_b.weight: Q5_K
31 blk.0.attn_sinks.weight: F32
32 blk.0.exp_probs_b.bias: F32
33 blk.0.exp_probs_b_vl.bias: F32
34 blk.0.ffn_down_exps.weight: IQ1_M
35 blk.0.ffn_down_shexp.weight: Q5_K
36 blk.0.ffn_gate_exps.weight: IQ1_M
37 blk.0.ffn_gate_inp.weight: BF16
38 blk.0.ffn_gate_shexp.weight: Q5_K
39 blk.0.ffn_norm.weight: F32
40 blk.0.ffn_up_exps.weight: IQ1_M
41 blk.0.ffn_up_shexp.weight: Q5_K
42 blk.0.hc_attn_base.weight: F32
43 blk.0.hc_attn_fn.weight: F32
44 blk.0.hc_attn_scale.weight: F32
45 blk.0.hc_ffn_base.weight: F32
46 blk.0.hc_ffn_fn.weight: F32
47 blk.0.hc_ffn_scale.weight: F32
48 blk.1.attn_kv.weight: Q8_0
49 blk.1.attn_kv_a_norm.weight: F32
50 blk.1.attn_norm.weight: F32
51 blk.1.attn_output_a.weight: Q5_K
52 blk.1.attn_output_b.weight: Q5_K
53 blk.1.attn_q_a.weight: Q8_0
54 blk.1.attn_q_a_norm.weight: F32
55 blk.1.attn_q_b.weight: Q5_K
56 blk.1.attn_sinks.weight: F32
57 blk.1.engram_gate.weight: F32
58 blk.1.engram_kv.weight: Q5_K
59 blk.1.exp_probs_b.bias: F32
60 blk.1.exp_probs_b_vl.bias: F32
61 blk.1.ffn_down_exps.weight: IQ1_M
62 blk.1.ffn_down_shexp.weight: Q5_K
63 blk.1.ffn_gate_exps.weight: IQ1_M
64 blk.1.ffn_gate_inp.weight: BF16
65 blk.1.ffn_gate_shexp.weight: Q5_K
66 blk.1.ffn_norm.weight: F32
67 blk.1.ffn_up_exps.weight: IQ1_M
68 blk.1.ffn_up_shexp.weight: Q5_K
69 blk.1.hc_attn_base.weight: F32
70 blk.1.hc_attn_fn.weight: F32
71 blk.1.hc_attn_scale.weight: F32
72 blk.1.hc_ffn_base.weight: F32
73 blk.1.hc_ffn_fn.weight: F32
74 blk.1.hc_ffn_scale.weight: F32
75 blk.2.attn_compressor_gate.weight: BF16
76 blk.2.attn_compressor_kv.weight: BF16
77 blk.2.attn_compressor_norm.weight: F32
78 blk.2.attn_kv.weight: Q8_0
79 blk.2.attn_kv_a_norm.weight: F32
80 blk.2.attn_norm.weight: F32
81 blk.2.attn_output_a.weight: Q5_K
82 blk.2.attn_output_b.weight: Q5_K
83 blk.2.attn_q_a.weight: Q8_0
84 blk.2.attn_q_a_norm.weight: F32
85 blk.2.attn_q_b.weight: Q5_K
86 blk.2.attn_sinks.weight: F32
87 blk.2.exp_probs_b.bias: F32
88 blk.2.exp_probs_b_vl.bias: F32
89 blk.2.ffn_down_exps.weight: IQ1_M
90 blk.2.ffn_down_shexp.weight: Q5_K
91 blk.2.ffn_gate_exps.weight: IQ1_M
92 blk.2.ffn_gate_inp.weight: BF16
93 blk.2.ffn_gate_shexp.weight: Q5_K
94 blk.2.ffn_norm.weight: F32
95 blk.2.ffn_up_exps.weight: IQ1_M
96 blk.2.ffn_up_shexp.weight: Q5_K
97 blk.2.hc_attn_base.weight: F32
98 blk.2.hc_attn_fn.weight: F32
99 blk.2.hc_attn_scale.weight: F32
100 blk.2.hc_ffn_base.weight: F32
101 blk.2.hc_ffn_fn.weight: F32
102 blk.2.hc_ffn_scale.weight: F32
103 blk.2.indexer.attn_k.weight: BF16
104 blk.2.indexer.attn_q_b.weight: BF16
105 blk.2.indexer.k_norm.weight: F32
106 blk.2.indexer.proj.weight: BF16
107 blk.3.attn_kv.weight: Q8_0
108 blk.3.attn_kv_a_norm.weight: F32
109 blk.3.attn_norm.weight: F32
110 blk.3.attn_output_a.weight: Q5_K
111 blk.3.attn_output_b.weight: Q5_K
112 blk.3.attn_q_a.weight: Q8_0
113 blk.3.attn_q_a_norm.weight: F32
114 blk.3.attn_q_b.weight: Q5_K
115 blk.3.attn_sinks.weight: F32
116 blk.3.exp_probs_b.bias: F32
117 blk.3.exp_probs_b_vl.bias: F32
118 blk.3.ffn_down_exps.weight: IQ1_M
119 blk.3.ffn_down_shexp.weight: Q5_K
120 blk.3.ffn_gate_exps.weight: IQ1_M
121 blk.3.ffn_gate_inp.weight: BF16
122 blk.3.ffn_gate_shexp.weight: Q5_K
123 blk.3.ffn_norm.weight: F32
124 blk.3.ffn_up_exps.weight: IQ1_M
125 blk.3.ffn_up_shexp.weight: Q5_K
126 blk.3.hc_attn_base.weight: F32
127 blk.3.hc_attn_fn.weight: F32
128 blk.3.hc_attn_scale.weight: F32
129 blk.3.hc_ffn_base.weight: F32
130 blk.3.hc_ffn_fn.weight: F32
131 blk.3.hc_ffn_scale.weight: F32
132 blk.4.attn_kv.weight: Q8_0
133 blk.4.attn_kv_a_norm.weight: F32
134 blk.4.attn_norm.weight: F32
135 blk.4.attn_output_a.weight: Q5_K
136 blk.4.attn_output_b.weight: Q5_K
137 blk.4.attn_q_a.weight: Q8_0
138 blk.4.attn_q_a_norm.weight: F32
139 blk.4.attn_q_b.weight: Q5_K
140 blk.4.attn_sinks.weight: F32
141 blk.4.exp_probs_b.bias: F32
142 blk.4.exp_probs_b_vl.bias: F32
143 blk.4.ffn_down_exps.weight: IQ1_M
144 blk.4.ffn_down_shexp.weight: Q5_K
145 blk.4.ffn_gate_exps.weight: IQ1_M
146 blk.4.ffn_gate_inp.weight: BF16
147 blk.4.ffn_gate_shexp.weight: Q5_K
148 blk.4.ffn_norm.weight: F32
149 blk.4.ffn_up_exps.weight: IQ1_M
150 blk.4.ffn_up_shexp.weight: Q5_K
151 blk.4.hc_attn_base.weight: F32
152 blk.4.hc_attn_fn.weight: F32
153 blk.4.hc_attn_scale.weight: F32
154 blk.4.hc_ffn_base.weight: F32
155 blk.4.hc_ffn_fn.weight: F32
156 blk.4.hc_ffn_scale.weight: F32
157 blk.5.attn_kv.weight: Q8_0
158 blk.5.attn_kv_a_norm.weight: F32
159 blk.5.attn_norm.weight: F32
160 blk.5.attn_output_a.weight: Q5_K
161 blk.5.attn_output_b.weight: Q5_K
162 blk.5.attn_q_a.weight: Q8_0
163 blk.5.attn_q_a_norm.weight: F32
164 blk.5.attn_q_b.weight: Q5_K
165 blk.5.attn_sinks.weight: F32
166 blk.5.exp_probs_b.bias: F32
167 blk.5.exp_probs_b_vl.bias: F32
168 blk.5.ffn_down_exps.weight: IQ1_M
169 blk.5.ffn_down_shexp.weight: Q5_K
170 blk.5.ffn_gate_exps.weight: IQ1_M
171 blk.5.ffn_gate_inp.weight: BF16
172 blk.5.ffn_gate_shexp.weight: Q5_K
173 blk.5.ffn_norm.weight: F32
174 blk.5.ffn_up_exps.weight: IQ1_M
175 blk.5.ffn_up_shexp.weight: Q5_K
176 blk.5.hc_attn_base.weight: F32
177 blk.5.hc_attn_fn.weight: F32
178 blk.5.hc_attn_scale.weight: F32
179 blk.5.hc_ffn_base.weight: F32
180 blk.5.hc_ffn_fn.weight: F32
181 blk.5.hc_ffn_scale.weight: F32
182 blk.6.attn_kv.weight: Q8_0
183 blk.6.attn_kv_a_norm.weight: F32
184 blk.6.attn_norm.weight: F32
185 blk.6.attn_output_a.weight: Q5_K
186 blk.6.attn_output_b.weight: Q5_K
187 blk.6.attn_q_a.weight: Q8_0
188 blk.6.attn_q_a_norm.weight: F32
189 blk.6.attn_q_b.weight: Q5_K
190 blk.6.attn_sinks.weight: F32
191 blk.6.exp_probs_b.bias: F32
192 blk.6.exp_probs_b_vl.bias: F32
193 blk.6.ffn_down_exps.weight: IQ1_M
194 blk.6.ffn_down_shexp.weight: Q5_K
195 blk.6.ffn_gate_exps.weight: IQ1_M
196 blk.6.ffn_gate_inp.weight: BF16
197 blk.6.ffn_gate_shexp.weight: Q5_K
198 blk.6.ffn_norm.weight: F32
199 blk.6.ffn_up_exps.weight: IQ1_M
200 blk.6.ffn_up_shexp.weight: Q5_K
201 blk.6.hc_attn_base.weight: F32
202 blk.6.hc_attn_fn.weight: F32
203 blk.6.hc_attn_scale.weight: F32
204 blk.6.hc_ffn_base.weight: F32
205 blk.6.hc_ffn_fn.weight: F32
206 blk.6.hc_ffn_scale.weight: F32
207 blk.7.attn_kv.weight: Q8_0
208 blk.7.attn_kv_a_norm.weight: F32
209 blk.7.attn_norm.weight: F32
210 blk.7.attn_output_a.weight: Q5_K
211 blk.7.attn_output_b.weight: Q5_K
212 blk.7.attn_q_a.weight: Q8_0
213 blk.7.attn_q_a_norm.weight: F32
214 blk.7.attn_q_b.weight: Q5_K
215 blk.7.attn_sinks.weight: F32
216 blk.7.exp_probs_b.bias: F32
217 blk.7.exp_probs_b_vl.bias: F32
218 blk.7.ffn_down_exps.weight: IQ2_XXS
219 blk.7.ffn_down_shexp.weight: Q5_K
220 blk.7.ffn_gate_exps.weight: IQ1_M
221 blk.7.ffn_gate_inp.weight: BF16
222 blk.7.ffn_gate_shexp.weight: Q5_K
223 blk.7.ffn_norm.weight: F32
224 blk.7.ffn_up_exps.weight: IQ1_M
225 blk.7.ffn_up_shexp.weight: Q5_K
226 blk.7.hc_attn_base.weight: F32
227 blk.7.hc_attn_fn.weight: F32
228 blk.7.hc_attn_scale.weight: F32
229 blk.7.hc_ffn_base.weight: F32
230 blk.7.hc_ffn_fn.weight: F32
231 blk.7.hc_ffn_scale.weight: F32
232 blk.8.attn_compressor_gate.weight: BF16
233 blk.8.attn_compressor_kv.weight: BF16
234 blk.8.attn_compressor_norm.weight: F32
235 blk.8.attn_kv.weight: Q8_0
236 blk.8.attn_kv_a_norm.weight: F32
237 blk.8.attn_norm.weight: F32
238 blk.8.attn_output_a.weight: Q5_K
239 blk.8.attn_output_b.weight: Q5_K
240 blk.8.attn_q_a.weight: Q8_0
241 blk.8.attn_q_a_norm.weight: F32
242 blk.8.attn_q_b.weight: Q5_K
243 blk.8.attn_sinks.weight: F32
244 blk.8.exp_probs_b.bias: F32
245 blk.8.exp_probs_b_vl.bias: F32
246 blk.8.ffn_down_exps.weight: IQ2_XXS
247 blk.8.ffn_down_shexp.weight: Q5_K
248 blk.8.ffn_gate_exps.weight: IQ1_M
249 blk.8.ffn_gate_inp.weight: BF16
250 blk.8.ffn_gate_shexp.weight: Q5_K
251 blk.8.ffn_norm.weight: F32
252 blk.8.ffn_up_exps.weight: IQ1_M
253 blk.8.ffn_up_shexp.weight: Q5_K
254 blk.8.hc_attn_base.weight: F32
255 blk.8.hc_attn_fn.weight: F32
256 blk.8.hc_attn_scale.weight: F32
257 blk.8.hc_ffn_base.weight: F32
258 blk.8.hc_ffn_fn.weight: F32
259 blk.8.hc_ffn_scale.weight: F32
260 blk.8.indexer.attn_k.weight: BF16
261 blk.8.indexer.attn_q_b.weight: BF16
262 blk.8.indexer.k_norm.weight: F32
263 blk.8.indexer.proj.weight: BF16
264 blk.9.attn_kv.weight: Q8_0
265 blk.9.attn_kv_a_norm.weight: F32
266 blk.9.attn_norm.weight: F32
267 blk.9.attn_output_a.weight: Q5_K
268 blk.9.attn_output_b.weight: Q5_K
269 blk.9.attn_q_a.weight: Q8_0
270 blk.9.attn_q_a_norm.weight: F32
271 blk.9.attn_q_b.weight: Q5_K
272 blk.9.attn_sinks.weight: F32
273 blk.9.exp_probs_b.bias: F32
274 blk.9.exp_probs_b_vl.bias: F32
275 blk.9.ffn_down_exps.weight: IQ1_M
276 blk.9.ffn_down_shexp.weight: Q5_K
277 blk.9.ffn_gate_exps.weight: IQ1_M
278 blk.9.ffn_gate_inp.weight: BF16
279 blk.9.ffn_gate_shexp.weight: Q5_K
280 blk.9.ffn_norm.weight: F32
281 blk.9.ffn_up_exps.weight: IQ1_M
282 blk.9.ffn_up_shexp.weight: Q5_K
283 blk.9.hc_attn_base.weight: F32
284 blk.9.hc_attn_fn.weight: F32
285 blk.9.hc_attn_scale.weight: F32
286 blk.9.hc_ffn_base.weight: F32
287 blk.9.hc_ffn_fn.weight: F32
288 blk.9.hc_ffn_scale.weight: F32
289 blk.10.attn_kv.weight: Q8_0
290 blk.10.attn_kv_a_norm.weight: F32
291 blk.10.attn_norm.weight: F32
292 blk.10.attn_output_a.weight: Q5_K
293 blk.10.attn_output_b.weight: Q5_K
294 blk.10.attn_q_a.weight: Q8_0
295 blk.10.attn_q_a_norm.weight: F32
296 blk.10.attn_q_b.weight: Q5_K
297 blk.10.attn_sinks.weight: F32
298 blk.10.exp_probs_b.bias: F32
299 blk.10.exp_probs_b_vl.bias: F32
300 blk.10.ffn_down_exps.weight: IQ2_XXS
301 blk.10.ffn_down_shexp.weight: Q5_K
302 blk.10.ffn_gate_exps.weight: IQ2_XXS
303 blk.10.ffn_gate_inp.weight: BF16
304 blk.10.ffn_gate_shexp.weight: Q5_K
305 blk.10.ffn_norm.weight: F32
306 blk.10.ffn_up_exps.weight: IQ2_XXS
307 blk.10.ffn_up_shexp.weight: Q5_K
308 blk.10.hc_attn_base.weight: F32
309 blk.10.hc_attn_fn.weight: F32
310 blk.10.hc_attn_scale.weight: F32
311 blk.10.hc_ffn_base.weight: F32
312 blk.10.hc_ffn_fn.weight: F32
313 blk.10.hc_ffn_scale.weight: F32
314 blk.11.attn_kv.weight: Q8_0
315 blk.11.attn_kv_a_norm.weight: F32
316 blk.11.attn_norm.weight: F32
317 blk.11.attn_output_a.weight: Q5_K
318 blk.11.attn_output_b.weight: Q5_K
319 blk.11.attn_q_a.weight: Q8_0
320 blk.11.attn_q_a_norm.weight: F32
321 blk.11.attn_q_b.weight: Q5_K
322 blk.11.attn_sinks.weight: F32
323 blk.11.exp_probs_b.bias: F32
324 blk.11.exp_probs_b_vl.bias: F32
325 blk.11.ffn_down_exps.weight: IQ1_M
326 blk.11.ffn_down_shexp.weight: Q5_K
327 blk.11.ffn_gate_exps.weight: IQ2_XXS
328 blk.11.ffn_gate_inp.weight: BF16
329 blk.11.ffn_gate_shexp.weight: Q5_K
330 blk.11.ffn_norm.weight: F32
331 blk.11.ffn_up_exps.weight: IQ2_XXS
332 blk.11.ffn_up_shexp.weight: Q5_K
333 blk.11.hc_attn_base.weight: F32
334 blk.11.hc_attn_fn.weight: F32
335 blk.11.hc_attn_scale.weight: F32
336 blk.11.hc_ffn_base.weight: F32
337 blk.11.hc_ffn_fn.weight: F32
338 blk.11.hc_ffn_scale.weight: F32
339 blk.12.attn_kv.weight: Q8_0
340 blk.12.attn_kv_a_norm.weight: F32
341 blk.12.attn_norm.weight: F32
342 blk.12.attn_output_a.weight: Q5_K
343 blk.12.attn_output_b.weight: Q5_K
344 blk.12.attn_q_a.weight: Q8_0
345 blk.12.attn_q_a_norm.weight: F32
346 blk.12.attn_q_b.weight: Q5_K
347 blk.12.attn_sinks.weight: F32
348 blk.12.exp_probs_b.bias: F32
349 blk.12.exp_probs_b_vl.bias: F32
350 blk.12.ffn_down_exps.weight: IQ1_M
351 blk.12.ffn_down_shexp.weight: Q5_K
352 blk.12.ffn_gate_exps.weight: IQ2_XXS
353 blk.12.ffn_gate_inp.weight: BF16
354 blk.12.ffn_gate_shexp.weight: Q5_K
355 blk.12.ffn_norm.weight: F32
356 blk.12.ffn_up_exps.weight: IQ2_XXS
357 blk.12.ffn_up_shexp.weight: Q5_K
358 blk.12.hc_attn_base.weight: F32
359 blk.12.hc_attn_fn.weight: F32
360 blk.12.hc_attn_scale.weight: F32
361 blk.12.hc_ffn_base.weight: F32
362 blk.12.hc_ffn_fn.weight: F32
363 blk.12.hc_ffn_scale.weight: F32
364 blk.13.attn_kv.weight: Q8_0
365 blk.13.attn_kv_a_norm.weight: F32
366 blk.13.attn_norm.weight: F32
367 blk.13.attn_output_a.weight: Q5_K
368 blk.13.attn_output_b.weight: Q5_K
369 blk.13.attn_q_a.weight: Q8_0
370 blk.13.attn_q_a_norm.weight: F32
371 blk.13.attn_q_b.weight: Q5_K
372 blk.13.attn_sinks.weight: F32
373 blk.13.exp_probs_b.bias: F32
374 blk.13.exp_probs_b_vl.bias: F32
375 blk.13.ffn_down_exps.weight: IQ1_M
376 blk.13.ffn_down_shexp.weight: Q5_K
377 blk.13.ffn_gate_exps.weight: IQ1_M
378 blk.13.ffn_gate_inp.weight: BF16
379 blk.13.ffn_gate_shexp.weight: Q5_K
380 blk.13.ffn_norm.weight: F32
381 blk.13.ffn_up_exps.weight: IQ1_M
382 blk.13.ffn_up_shexp.weight: Q5_K
383 blk.13.hc_attn_base.weight: F32
384 blk.13.hc_attn_fn.weight: F32
385 blk.13.hc_attn_scale.weight: F32
386 blk.13.hc_ffn_base.weight: F32
387 blk.13.hc_ffn_fn.weight: F32
388 blk.13.hc_ffn_scale.weight: F32
389 blk.14.attn_compressor_gate.weight: BF16
390 blk.14.attn_compressor_kv.weight: BF16
391 blk.14.attn_compressor_norm.weight: F32
392 blk.14.attn_kv.weight: Q8_0
393 blk.14.attn_kv_a_norm.weight: F32
394 blk.14.attn_norm.weight: F32
395 blk.14.attn_output_a.weight: Q5_K
396 blk.14.attn_output_b.weight: Q5_K
397 blk.14.attn_q_a.weight: Q8_0
398 blk.14.attn_q_a_norm.weight: F32
399 blk.14.attn_q_b.weight: Q5_K
400 blk.14.attn_sinks.weight: F32
401 blk.14.engram_gate.weight: F32
402 blk.14.engram_kv.weight: Q5_K
403 blk.14.exp_probs_b.bias: F32
404 blk.14.exp_probs_b_vl.bias: F32
405 blk.14.ffn_down_exps.weight: IQ1_M
406 blk.14.ffn_down_shexp.weight: Q5_K
407 blk.14.ffn_gate_exps.weight: IQ1_M
408 blk.14.ffn_gate_inp.weight: BF16
409 blk.14.ffn_gate_shexp.weight: Q5_K
410 blk.14.ffn_norm.weight: F32
411 blk.14.ffn_up_exps.weight: IQ1_M
412 blk.14.ffn_up_shexp.weight: Q5_K
413 blk.14.hc_attn_base.weight: F32
414 blk.14.hc_attn_fn.weight: F32
415 blk.14.hc_attn_scale.weight: F32
416 blk.14.hc_ffn_base.weight: F32
417 blk.14.hc_ffn_fn.weight: F32
418 blk.14.hc_ffn_scale.weight: F32
419 blk.14.indexer.attn_k.weight: BF16
420 blk.14.indexer.attn_q_b.weight: BF16
421 blk.14.indexer.k_norm.weight: F32
422 blk.14.indexer.proj.weight: BF16
423 blk.15.attn_kv.weight: Q8_0
424 blk.15.attn_kv_a_norm.weight: F32
425 blk.15.attn_norm.weight: F32
426 blk.15.attn_output_a.weight: Q5_K
427 blk.15.attn_output_b.weight: Q5_K
428 blk.15.attn_q_a.weight: Q8_0
429 blk.15.attn_q_a_norm.weight: F32
430 blk.15.attn_q_b.weight: Q5_K
431 blk.15.attn_sinks.weight: F32
432 blk.15.exp_probs_b.bias: F32
433 blk.15.exp_probs_b_vl.bias: F32
434 blk.15.ffn_down_exps.weight: IQ1_M
435 blk.15.ffn_down_shexp.weight: Q5_K
436 blk.15.ffn_gate_exps.weight: IQ2_XXS
437 blk.15.ffn_gate_inp.weight: BF16
438 blk.15.ffn_gate_shexp.weight: Q5_K
439 blk.15.ffn_norm.weight: F32
440 blk.15.ffn_up_exps.weight: IQ2_XXS
441 blk.15.ffn_up_shexp.weight: Q5_K
442 blk.15.hc_attn_base.weight: F32
443 blk.15.hc_attn_fn.weight: F32
444 blk.15.hc_attn_scale.weight: F32
445 blk.15.hc_ffn_base.weight: F32
446 blk.15.hc_ffn_fn.weight: F32
447 blk.15.hc_ffn_scale.weight: F32
448 blk.16.attn_kv.weight: Q8_0
449 blk.16.attn_kv_a_norm.weight: F32
450 blk.16.attn_norm.weight: F32
451 blk.16.attn_output_a.weight: Q5_K
452 blk.16.attn_output_b.weight: Q5_K
453 blk.16.attn_q_a.weight: Q8_0
454 blk.16.attn_q_a_norm.weight: F32
455 blk.16.attn_q_b.weight: Q5_K
456 blk.16.attn_sinks.weight: F32
457 blk.16.exp_probs_b.bias: F32
458 blk.16.exp_probs_b_vl.bias: F32
459 blk.16.ffn_down_exps.weight: IQ1_M
460 blk.16.ffn_down_shexp.weight: Q5_K
461 blk.16.ffn_gate_exps.weight: IQ1_M
462 blk.16.ffn_gate_inp.weight: BF16
463 blk.16.ffn_gate_shexp.weight: Q5_K
464 blk.16.ffn_norm.weight: F32
465 blk.16.ffn_up_exps.weight: IQ1_M
466 blk.16.ffn_up_shexp.weight: Q5_K
467 blk.16.hc_attn_base.weight: F32
468 blk.16.hc_attn_fn.weight: F32
469 blk.16.hc_attn_scale.weight: F32
470 blk.16.hc_ffn_base.weight: F32
471 blk.16.hc_ffn_fn.weight: F32
472 blk.16.hc_ffn_scale.weight: F32
473 blk.17.attn_kv.weight: Q8_0
474 blk.17.attn_kv_a_norm.weight: F32
475 blk.17.attn_norm.weight: F32
476 blk.17.attn_output_a.weight: Q5_K
477 blk.17.attn_output_b.weight: Q5_K
478 blk.17.attn_q_a.weight: Q8_0
479 blk.17.attn_q_a_norm.weight: F32
480 blk.17.attn_q_b.weight: Q5_K
481 blk.17.attn_sinks.weight: F32
482 blk.17.exp_probs_b.bias: F32
483 blk.17.exp_probs_b_vl.bias: F32
484 blk.17.ffn_down_exps.weight: IQ1_M
485 blk.17.ffn_down_shexp.weight: Q5_K
486 blk.17.ffn_gate_exps.weight: IQ2_XS
487 blk.17.ffn_gate_inp.weight: BF16
488 blk.17.ffn_gate_shexp.weight: Q5_K
489 blk.17.ffn_norm.weight: F32
490 blk.17.ffn_up_exps.weight: IQ2_XS
491 blk.17.ffn_up_shexp.weight: Q5_K
492 blk.17.hc_attn_base.weight: F32
493 blk.17.hc_attn_fn.weight: F32
494 blk.17.hc_attn_scale.weight: F32
495 blk.17.hc_ffn_base.weight: F32
496 blk.17.hc_ffn_fn.weight: F32
497 blk.17.hc_ffn_scale.weight: F32
498 blk.18.attn_kv.weight: Q8_0
499 blk.18.attn_kv_a_norm.weight: F32
500 blk.18.attn_norm.weight: F32
501 blk.18.attn_output_a.weight: Q5_K
502 blk.18.attn_output_b.weight: Q5_K
503 blk.18.attn_q_a.weight: Q8_0
504 blk.18.attn_q_a_norm.weight: F32
505 blk.18.attn_q_b.weight: Q5_K
506 blk.18.attn_sinks.weight: F32
507 blk.18.exp_probs_b.bias: F32
508 blk.18.exp_probs_b_vl.bias: F32
509 blk.18.ffn_down_exps.weight: IQ2_XXS
510 blk.18.ffn_down_shexp.weight: Q5_K
511 blk.18.ffn_gate_exps.weight: IQ2_XXS
512 blk.18.ffn_gate_inp.weight: BF16
513 blk.18.ffn_gate_shexp.weight: Q5_K
514 blk.18.ffn_norm.weight: F32
515 blk.18.ffn_up_exps.weight: IQ2_XXS
516 blk.18.ffn_up_shexp.weight: Q5_K
517 blk.18.hc_attn_base.weight: F32
518 blk.18.hc_attn_fn.weight: F32
519 blk.18.hc_attn_scale.weight: F32
520 blk.18.hc_ffn_base.weight: F32
521 blk.18.hc_ffn_fn.weight: F32
522 blk.18.hc_ffn_scale.weight: F32
523 blk.19.attn_kv.weight: Q8_0
524 blk.19.attn_kv_a_norm.weight: F32
525 blk.19.attn_norm.weight: F32
526 blk.19.attn_output_a.weight: Q5_K
527 blk.19.attn_output_b.weight: Q5_K
528 blk.19.attn_q_a.weight: Q8_0
529 blk.19.attn_q_a_norm.weight: F32
530 blk.19.attn_q_b.weight: Q5_K
531 blk.19.attn_sinks.weight: F32
532 blk.19.exp_probs_b.bias: F32
533 blk.19.exp_probs_b_vl.bias: F32
534 blk.19.ffn_down_exps.weight: IQ1_M
535 blk.19.ffn_down_shexp.weight: Q5_K
536 blk.19.ffn_gate_exps.weight: IQ2_XS
537 blk.19.ffn_gate_inp.weight: BF16
538 blk.19.ffn_gate_shexp.weight: Q5_K
539 blk.19.ffn_norm.weight: F32
540 blk.19.ffn_up_exps.weight: IQ2_XS
541 blk.19.ffn_up_shexp.weight: Q5_K
542 blk.19.hc_attn_base.weight: F32
543 blk.19.hc_attn_fn.weight: F32
544 blk.19.hc_attn_scale.weight: F32
545 blk.19.hc_ffn_base.weight: F32
546 blk.19.hc_ffn_fn.weight: F32
547 blk.19.hc_ffn_scale.weight: F32
548 blk.20.attn_compressor_kv.weight: BF16
549 blk.20.attn_compressor_norm.weight: F32
550 blk.20.attn_kv.weight: Q8_0
551 blk.20.attn_kv_a_norm.weight: F32
552 blk.20.attn_norm.weight: F32
553 blk.20.attn_output_a.weight: Q5_K
554 blk.20.attn_output_b.weight: Q5_K
555 blk.20.attn_q_a.weight: Q8_0
556 blk.20.attn_q_a_norm.weight: F32
557 blk.20.attn_q_b.weight: Q5_K
558 blk.20.attn_sinks.weight: F32
559 blk.20.exp_probs_b.bias: F32
560 blk.20.exp_probs_b_vl.bias: F32
561 blk.20.ffn_down_exps.weight: IQ1_M
562 blk.20.ffn_down_shexp.weight: Q5_K
563 blk.20.ffn_gate_exps.weight: IQ1_M
564 blk.20.ffn_gate_inp.weight: BF16
565 blk.20.ffn_gate_shexp.weight: Q5_K
566 blk.20.ffn_norm.weight: F32
567 blk.20.ffn_up_exps.weight: IQ1_M
568 blk.20.ffn_up_shexp.weight: Q5_K
569 blk.20.hc_attn_base.weight: F32
570 blk.20.hc_attn_fn.weight: F32
571 blk.20.hc_attn_scale.weight: F32
572 blk.20.hc_ffn_base.weight: F32
573 blk.20.hc_ffn_fn.weight: F32
574 blk.20.hc_ffn_scale.weight: F32
575 blk.20.indexer.attn_k.weight: BF16
576 blk.20.indexer.attn_q_b.weight: BF16
577 blk.20.indexer.k_norm.weight: F32
578 blk.20.indexer.proj.weight: BF16
579 blk.21.attn_kv.weight: Q8_0
580 blk.21.attn_kv_a_norm.weight: F32
581 blk.21.attn_norm.weight: F32
582 blk.21.attn_output_a.weight: Q5_K
583 blk.21.attn_output_b.weight: Q5_K
584 blk.21.attn_q_a.weight: Q8_0
585 blk.21.attn_q_a_norm.weight: F32
586 blk.21.attn_q_b.weight: Q5_K
587 blk.21.attn_sinks.weight: F32
588 blk.21.exp_probs_b.bias: F32
589 blk.21.exp_probs_b_vl.bias: F32
590 blk.21.ffn_down_exps.weight: IQ1_M
591 blk.21.ffn_down_shexp.weight: Q5_K
592 blk.21.ffn_gate_exps.weight: IQ1_M
593 blk.21.ffn_gate_inp.weight: BF16
594 blk.21.ffn_gate_shexp.weight: Q5_K
595 blk.21.ffn_norm.weight: F32
596 blk.21.ffn_up_exps.weight: IQ1_M
597 blk.21.ffn_up_shexp.weight: Q5_K
598 blk.21.hc_attn_base.weight: F32
599 blk.21.hc_attn_fn.weight: F32
600 blk.21.hc_attn_scale.weight: F32
601 blk.21.hc_ffn_base.weight: F32
602 blk.21.hc_ffn_fn.weight: F32
603 blk.21.hc_ffn_scale.weight: F32
604 blk.22.attn_kv.weight: Q8_0
605 blk.22.attn_kv_a_norm.weight: F32
606 blk.22.attn_norm.weight: F32
607 blk.22.attn_output_a.weight: Q5_K
608 blk.22.attn_output_b.weight: Q5_K
609 blk.22.attn_q_a.weight: Q8_0
610 blk.22.attn_q_a_norm.weight: F32
611 blk.22.attn_q_b.weight: Q5_K
612 blk.22.attn_sinks.weight: F32
613 blk.22.exp_probs_b.bias: F32
614 blk.22.exp_probs_b_vl.bias: F32
615 blk.22.ffn_down_exps.weight: IQ2_XXS
616 blk.22.ffn_down_shexp.weight: Q5_K
617 blk.22.ffn_gate_exps.weight: IQ2_XXS
618 blk.22.ffn_gate_inp.weight: BF16
619 blk.22.ffn_gate_shexp.weight: Q5_K
620 blk.22.ffn_norm.weight: F32
621 blk.22.ffn_up_exps.weight: IQ2_XXS
622 blk.22.ffn_up_shexp.weight: Q5_K
623 blk.22.hc_attn_base.weight: F32
624 blk.22.hc_attn_fn.weight: F32
625 blk.22.hc_attn_scale.weight: F32
626 blk.22.hc_ffn_base.weight: F32
627 blk.22.hc_ffn_fn.weight: F32
628 blk.22.hc_ffn_scale.weight: F32
629 blk.23.attn_kv.weight: Q8_0
630 blk.23.attn_kv_a_norm.weight: F32
631 blk.23.attn_norm.weight: F32
632 blk.23.attn_output_a.weight: Q5_K
633 blk.23.attn_output_b.weight: Q5_K
634 blk.23.attn_q_a.weight: Q8_0
635 blk.23.attn_q_a_norm.weight: F32
636 blk.23.attn_q_b.weight: Q5_K
637 blk.23.attn_sinks.weight: F32
638 blk.23.exp_probs_b.bias: F32
639 blk.23.exp_probs_b_vl.bias: F32
640 blk.23.ffn_down_exps.weight: IQ2_XS
641 blk.23.ffn_down_shexp.weight: Q5_K
642 blk.23.ffn_gate_exps.weight: IQ2_XXS
643 blk.23.ffn_gate_inp.weight: BF16
644 blk.23.ffn_gate_shexp.weight: Q5_K
645 blk.23.ffn_norm.weight: F32
646 blk.23.ffn_up_exps.weight: IQ2_XXS
647 blk.23.ffn_up_shexp.weight: Q5_K
648 blk.23.hc_attn_base.weight: F32
649 blk.23.hc_attn_fn.weight: F32
650 blk.23.hc_attn_scale.weight: F32
651 blk.23.hc_ffn_base.weight: F32
652 blk.23.hc_ffn_fn.weight: F32
653 blk.23.hc_ffn_scale.weight: F32
654 blk.24.attn_kv.weight: Q8_0
655 blk.24.attn_kv_a_norm.weight: F32
656 blk.24.attn_norm.weight: F32
657 blk.24.attn_output_a.weight: Q5_K
658 blk.24.attn_output_b.weight: Q5_K
659 blk.24.attn_q_a.weight: Q8_0
660 blk.24.attn_q_a_norm.weight: F32
661 blk.24.attn_q_b.weight: Q5_K
662 blk.24.attn_sinks.weight: F32
663 blk.24.exp_probs_b.bias: F32
664 blk.24.exp_probs_b_vl.bias: F32
665 blk.24.ffn_down_exps.weight: IQ1_M
666 blk.24.ffn_down_shexp.weight: Q5_K
667 blk.24.ffn_gate_exps.weight: IQ2_XXS
668 blk.24.ffn_gate_inp.weight: BF16
669 blk.24.ffn_gate_shexp.weight: Q5_K
670 blk.24.ffn_norm.weight: F32
671 blk.24.ffn_up_exps.weight: IQ2_XXS
672 blk.24.ffn_up_shexp.weight: Q5_K
673 blk.24.hc_attn_base.weight: F32
674 blk.24.hc_attn_fn.weight: F32
675 blk.24.hc_attn_scale.weight: F32
676 blk.24.hc_ffn_base.weight: F32
677 blk.24.hc_ffn_fn.weight: F32
678 blk.24.hc_ffn_scale.weight: F32
679 blk.24.indexer.attn_q_b.weight: BF16
680 blk.24.indexer.proj.weight: BF16
681 blk.25.attn_kv.weight: Q8_0
682 blk.25.attn_kv_a_norm.weight: F32
683 blk.25.attn_norm.weight: F32
684 blk.25.attn_output_a.weight: Q5_K
685 blk.25.attn_output_b.weight: Q5_K
686 blk.25.attn_q_a.weight: Q8_0
687 blk.25.attn_q_a_norm.weight: F32
688 blk.25.attn_q_b.weight: Q5_K
689 blk.25.attn_sinks.weight: F32
690 blk.25.exp_probs_b.bias: F32
691 blk.25.exp_probs_b_vl.bias: F32
692 blk.25.ffn_down_exps.weight: IQ2_XXS
693 blk.25.ffn_down_shexp.weight: Q5_K
694 blk.25.ffn_gate_exps.weight: IQ2_XXS
695 blk.25.ffn_gate_inp.weight: BF16
696 blk.25.ffn_gate_shexp.weight: Q5_K
697 blk.25.ffn_norm.weight: F32
698 blk.25.ffn_up_exps.weight: IQ2_XXS
699 blk.25.ffn_up_shexp.weight: Q5_K
700 blk.25.hc_attn_base.weight: F32
701 blk.25.hc_attn_fn.weight: F32
702 blk.25.hc_attn_scale.weight: F32
703 blk.25.hc_ffn_base.weight: F32
704 blk.25.hc_ffn_fn.weight: F32
705 blk.25.hc_ffn_scale.weight: F32
706 blk.26.attn_kv.weight: Q8_0
707 blk.26.attn_kv_a_norm.weight: F32
708 blk.26.attn_norm.weight: F32
709 blk.26.attn_output_a.weight: Q5_K
710 blk.26.attn_output_b.weight: Q5_K
711 blk.26.attn_q_a.weight: Q8_0
712 blk.26.attn_q_a_norm.weight: F32
713 blk.26.attn_q_b.weight: Q5_K
714 blk.26.attn_sinks.weight: F32
715 blk.26.exp_probs_b.bias: F32
716 blk.26.exp_probs_b_vl.bias: F32
717 blk.26.ffn_down_exps.weight: IQ2_XS
718 blk.26.ffn_down_shexp.weight: Q5_K
719 blk.26.ffn_gate_exps.weight: IQ2_XXS
720 blk.26.ffn_gate_inp.weight: BF16
721 blk.26.ffn_gate_shexp.weight: Q5_K
722 blk.26.ffn_norm.weight: F32
723 blk.26.ffn_up_exps.weight: IQ2_XXS
724 blk.26.ffn_up_shexp.weight: Q5_K
725 blk.26.hc_attn_base.weight: F32
726 blk.26.hc_attn_fn.weight: F32
727 blk.26.hc_attn_scale.weight: F32
728 blk.26.hc_ffn_base.weight: F32
729 blk.26.hc_ffn_fn.weight: F32
730 blk.26.hc_ffn_scale.weight: F32
731 blk.27.attn_kv.weight: Q8_0
732 blk.27.attn_kv_a_norm.weight: F32
733 blk.27.attn_norm.weight: F32
734 blk.27.attn_output_a.weight: Q5_K
735 blk.27.attn_output_b.weight: Q5_K
736 blk.27.attn_q_a.weight: Q8_0
737 blk.27.attn_q_a_norm.weight: F32
738 blk.27.attn_q_b.weight: Q5_K
739 blk.27.attn_sinks.weight: F32
740 blk.27.exp_probs_b.bias: F32
741 blk.27.exp_probs_b_vl.bias: F32
742 blk.27.ffn_down_exps.weight: IQ1_M
743 blk.27.ffn_down_shexp.weight: Q5_K
744 blk.27.ffn_gate_exps.weight: IQ2_XS
745 blk.27.ffn_gate_inp.weight: BF16
746 blk.27.ffn_gate_shexp.weight: Q5_K
747 blk.27.ffn_norm.weight: F32
748 blk.27.ffn_up_exps.weight: IQ2_XS
749 blk.27.ffn_up_shexp.weight: Q5_K
750 blk.27.hc_attn_base.weight: F32
751 blk.27.hc_attn_fn.weight: F32
752 blk.27.hc_attn_scale.weight: F32
753 blk.27.hc_ffn_base.weight: F32
754 blk.27.hc_ffn_fn.weight: F32
755 blk.27.hc_ffn_scale.weight: F32
756 blk.28.attn_kv.weight: Q8_0
757 blk.28.attn_kv_a_norm.weight: F32
758 blk.28.attn_norm.weight: F32
759 blk.28.attn_output_a.weight: Q5_K
760 blk.28.attn_output_b.weight: Q5_K
761 blk.28.attn_q_a.weight: Q8_0
762 blk.28.attn_q_a_norm.weight: F32
763 blk.28.attn_q_b.weight: Q5_K
764 blk.28.attn_sinks.weight: F32
765 blk.28.exp_probs_b.bias: F32
766 blk.28.exp_probs_b_vl.bias: F32
767 blk.28.ffn_down_exps.weight: IQ2_XXS
768 blk.28.ffn_down_shexp.weight: Q5_K
769 blk.28.ffn_gate_exps.weight: IQ2_XXS
770 blk.28.ffn_gate_inp.weight: BF16
771 blk.28.ffn_gate_shexp.weight: Q5_K
772 blk.28.ffn_norm.weight: F32
773 blk.28.ffn_up_exps.weight: IQ2_XXS
774 blk.28.ffn_up_shexp.weight: Q5_K
775 blk.28.hc_attn_base.weight: F32
776 blk.28.hc_attn_fn.weight: F32
777 blk.28.hc_attn_scale.weight: F32
778 blk.28.hc_ffn_base.weight: F32
779 blk.28.hc_ffn_fn.weight: F32
780 blk.28.hc_ffn_scale.weight: F32
781 blk.28.indexer.attn_q_b.weight: BF16
782 blk.28.indexer.proj.weight: BF16
783 blk.29.attn_kv.weight: Q8_0
784 blk.29.attn_kv_a_norm.weight: F32
785 blk.29.attn_norm.weight: F32
786 blk.29.attn_output_a.weight: Q5_K
787 blk.29.attn_output_b.weight: Q5_K
788 blk.29.attn_q_a.weight: Q8_0
789 blk.29.attn_q_a_norm.weight: F32
790 blk.29.attn_q_b.weight: Q5_K
791 blk.29.attn_sinks.weight: F32
792 blk.29.exp_probs_b.bias: F32
793 blk.29.exp_probs_b_vl.bias: F32
794 blk.29.ffn_down_exps.weight: IQ1_M
795 blk.29.ffn_down_shexp.weight: Q5_K
796 blk.29.ffn_gate_exps.weight: IQ2_XXS
797 blk.29.ffn_gate_inp.weight: BF16
798 blk.29.ffn_gate_shexp.weight: Q5_K
799 blk.29.ffn_norm.weight: F32
800 blk.29.ffn_up_exps.weight: IQ2_XXS
801 blk.29.ffn_up_shexp.weight: Q5_K
802 blk.29.hc_attn_base.weight: F32
803 blk.29.hc_attn_fn.weight: F32
804 blk.29.hc_attn_scale.weight: F32
805 blk.29.hc_ffn_base.weight: F32
806 blk.29.hc_ffn_fn.weight: F32
807 blk.29.hc_ffn_scale.weight: F32
808 blk.30.attn_kv.weight: Q8_0
809 blk.30.attn_kv_a_norm.weight: F32
810 blk.30.attn_norm.weight: F32
811 blk.30.attn_output_a.weight: Q5_K
812 blk.30.attn_output_b.weight: Q5_K
813 blk.30.attn_q_a.weight: Q8_0
814 blk.30.attn_q_a_norm.weight: F32
815 blk.30.attn_q_b.weight: Q5_K
816 blk.30.attn_sinks.weight: F32
817 blk.30.exp_probs_b.bias: F32
818 blk.30.exp_probs_b_vl.bias: F32
819 blk.30.ffn_down_exps.weight: IQ1_M
820 blk.30.ffn_down_shexp.weight: Q5_K
821 blk.30.ffn_gate_exps.weight: IQ1_M
822 blk.30.ffn_gate_inp.weight: BF16
823 blk.30.ffn_gate_shexp.weight: Q5_K
824 blk.30.ffn_norm.weight: F32
825 blk.30.ffn_up_exps.weight: IQ1_M
826 blk.30.ffn_up_shexp.weight: Q5_K
827 blk.30.hc_attn_base.weight: F32
828 blk.30.hc_attn_fn.weight: F32
829 blk.30.hc_attn_scale.weight: F32
830 blk.30.hc_ffn_base.weight: F32
831 blk.30.hc_ffn_fn.weight: F32
832 blk.30.hc_ffn_scale.weight: F32
833 blk.31.attn_kv.weight: Q8_0
834 blk.31.attn_kv_a_norm.weight: F32
835 blk.31.attn_norm.weight: F32
836 blk.31.attn_output_a.weight: Q5_K
837 blk.31.attn_output_b.weight: Q5_K
838 blk.31.attn_q_a.weight: Q8_0
839 blk.31.attn_q_a_norm.weight: F32
840 blk.31.attn_q_b.weight: Q5_K
841 blk.31.attn_sinks.weight: F32
842 blk.31.exp_probs_b.bias: F32
843 blk.31.exp_probs_b_vl.bias: F32
844 blk.31.ffn_down_exps.weight: IQ1_M
845 blk.31.ffn_down_shexp.weight: Q5_K
846 blk.31.ffn_gate_exps.weight: IQ2_XXS
847 blk.31.ffn_gate_inp.weight: BF16
848 blk.31.ffn_gate_shexp.weight: Q5_K
849 blk.31.ffn_norm.weight: F32
850 blk.31.ffn_up_exps.weight: IQ2_XXS
851 blk.31.ffn_up_shexp.weight: Q5_K
852 blk.31.hc_attn_base.weight: F32
853 blk.31.hc_attn_fn.weight: F32
854 blk.31.hc_attn_scale.weight: F32
855 blk.31.hc_ffn_base.weight: F32
856 blk.31.hc_ffn_fn.weight: F32
857 blk.31.hc_ffn_scale.weight: F32
858 blk.32.attn_kv.weight: Q8_0
859 blk.32.attn_kv_a_norm.weight: F32
860 blk.32.attn_norm.weight: F32
861 blk.32.attn_output_a.weight: Q5_K
862 blk.32.attn_output_b.weight: Q5_K
863 blk.32.attn_q_a.weight: Q8_0
864 blk.32.attn_q_a_norm.weight: F32
865 blk.32.attn_q_b.weight: Q5_K
866 blk.32.attn_sinks.weight: F32
867 blk.32.exp_probs_b.bias: F32
868 blk.32.exp_probs_b_vl.bias: F32
869 blk.32.ffn_down_exps.weight: IQ1_M
870 blk.32.ffn_down_shexp.weight: Q5_K
871 blk.32.ffn_gate_exps.weight: IQ2_XXS
872 blk.32.ffn_gate_inp.weight: BF16
873 blk.32.ffn_gate_shexp.weight: Q5_K
874 blk.32.ffn_norm.weight: F32
875 blk.32.ffn_up_exps.weight: IQ2_XXS
876 blk.32.ffn_up_shexp.weight: Q5_K
877 blk.32.hc_attn_base.weight: F32
878 blk.32.hc_attn_fn.weight: F32
879 blk.32.hc_attn_scale.weight: F32
880 blk.32.hc_ffn_base.weight: F32
881 blk.32.hc_ffn_fn.weight: F32
882 blk.32.hc_ffn_scale.weight: F32
883 blk.32.indexer.attn_q_b.weight: BF16
884 blk.32.indexer.proj.weight: BF16
885 blk.33.attn_kv.weight: Q8_0
886 blk.33.attn_kv_a_norm.weight: F32
887 blk.33.attn_norm.weight: F32
888 blk.33.attn_output_a.weight: Q5_K
889 blk.33.attn_output_b.weight: Q5_K
890 blk.33.attn_q_a.weight: Q8_0
891 blk.33.attn_q_a_norm.weight: F32
892 blk.33.attn_q_b.weight: Q5_K
893 blk.33.attn_sinks.weight: F32
894 blk.33.exp_probs_b.bias: F32
895 blk.33.exp_probs_b_vl.bias: F32
896 blk.33.ffn_down_exps.weight: IQ2_XS
897 blk.33.ffn_down_shexp.weight: Q5_K
898 blk.33.ffn_gate_exps.weight: IQ2_XXS
899 blk.33.ffn_gate_inp.weight: BF16
900 blk.33.ffn_gate_shexp.weight: Q5_K
901 blk.33.ffn_norm.weight: F32
902 blk.33.ffn_up_exps.weight: IQ2_XXS
903 blk.33.ffn_up_shexp.weight: Q5_K
904 blk.33.hc_attn_base.weight: F32
905 blk.33.hc_attn_fn.weight: F32
906 blk.33.hc_attn_scale.weight: F32
907 blk.33.hc_ffn_base.weight: F32
908 blk.33.hc_ffn_fn.weight: F32
909 blk.33.hc_ffn_scale.weight: F32
910 blk.34.attn_kv.weight: Q8_0
911 blk.34.attn_kv_a_norm.weight: F32
912 blk.34.attn_norm.weight: F32
913 blk.34.attn_output_a.weight: Q5_K
914 blk.34.attn_output_b.weight: Q5_K
915 blk.34.attn_q_a.weight: Q8_0
916 blk.34.attn_q_a_norm.weight: F32
917 blk.34.attn_q_b.weight: Q5_K
918 blk.34.attn_sinks.weight: F32
919 blk.34.exp_probs_b.bias: F32
920 blk.34.exp_probs_b_vl.bias: F32
921 blk.34.ffn_down_exps.weight: IQ1_M
922 blk.34.ffn_down_shexp.weight: Q5_K
923 blk.34.ffn_gate_exps.weight: IQ2_XS
924 blk.34.ffn_gate_inp.weight: BF16
925 blk.34.ffn_gate_shexp.weight: Q5_K
926 blk.34.ffn_norm.weight: F32
927 blk.34.ffn_up_exps.weight: IQ2_XS
928 blk.34.ffn_up_shexp.weight: Q5_K
929 blk.34.hc_attn_base.weight: F32
930 blk.34.hc_attn_fn.weight: F32
931 blk.34.hc_attn_scale.weight: F32
932 blk.34.hc_ffn_base.weight: F32
933 blk.34.hc_ffn_fn.weight: F32
934 blk.34.hc_ffn_scale.weight: F32
935 blk.35.attn_kv.weight: Q8_0
936 blk.35.attn_kv_a_norm.weight: F32
937 blk.35.attn_norm.weight: F32
938 blk.35.attn_output_a.weight: Q5_K
939 blk.35.attn_output_b.weight: Q5_K
940 blk.35.attn_q_a.weight: Q8_0
941 blk.35.attn_q_a_norm.weight: F32
942 blk.35.attn_q_b.weight: Q5_K
943 blk.35.attn_sinks.weight: F32
944 blk.35.exp_probs_b.bias: F32
945 blk.35.exp_probs_b_vl.bias: F32
946 blk.35.ffn_down_exps.weight: IQ1_M
947 blk.35.ffn_down_shexp.weight: Q5_K
948 blk.35.ffn_gate_exps.weight: IQ2_XS
949 blk.35.ffn_gate_inp.weight: BF16
950 blk.35.ffn_gate_shexp.weight: Q5_K
951 blk.35.ffn_norm.weight: F32
952 blk.35.ffn_up_exps.weight: IQ2_XS
953 blk.35.ffn_up_shexp.weight: Q5_K
954 blk.35.hc_attn_base.weight: F32
955 blk.35.hc_attn_fn.weight: F32
956 blk.35.hc_attn_scale.weight: F32
957 blk.35.hc_ffn_base.weight: F32
958 blk.35.hc_ffn_fn.weight: F32
959 blk.35.hc_ffn_scale.weight: F32
960 blk.36.attn_kv.weight: Q8_0
961 blk.36.attn_kv_a_norm.weight: F32
962 blk.36.attn_norm.weight: F32
963 blk.36.attn_output_a.weight: Q5_K
964 blk.36.attn_output_b.weight: Q5_K
965 blk.36.attn_q_a.weight: Q8_0
966 blk.36.attn_q_a_norm.weight: F32
967 blk.36.attn_q_b.weight: Q5_K
968 blk.36.attn_sinks.weight: F32
969 blk.36.exp_probs_b.bias: F32
970 blk.36.exp_probs_b_vl.bias: F32
971 blk.36.ffn_down_exps.weight: IQ1_M
972 blk.36.ffn_down_shexp.weight: Q5_K
973 blk.36.ffn_gate_exps.weight: IQ2_XXS
974 blk.36.ffn_gate_inp.weight: BF16
975 blk.36.ffn_gate_shexp.weight: Q5_K
976 blk.36.ffn_norm.weight: F32
977 blk.36.ffn_up_exps.weight: IQ2_XXS
978 blk.36.ffn_up_shexp.weight: Q5_K
979 blk.36.hc_attn_base.weight: F32
980 blk.36.hc_attn_fn.weight: F32
981 blk.36.hc_attn_scale.weight: F32
982 blk.36.hc_ffn_base.weight: F32
983 blk.36.hc_ffn_fn.weight: F32
984 blk.36.hc_ffn_scale.weight: F32
985 blk.36.indexer.attn_q_b.weight: BF16
986 blk.36.indexer.proj.weight: BF16
987 blk.37.attn_kv.weight: Q8_0
988 blk.37.attn_kv_a_norm.weight: F32
989 blk.37.attn_norm.weight: F32
990 blk.37.attn_output_a.weight: Q5_K
991 blk.37.attn_output_b.weight: Q5_K
992 blk.37.attn_q_a.weight: Q8_0
993 blk.37.attn_q_a_norm.weight: F32
994 blk.37.attn_q_b.weight: Q5_K
995 blk.37.attn_sinks.weight: F32
996 blk.37.exp_probs_b.bias: F32
997 blk.37.exp_probs_b_vl.bias: F32
998 blk.37.ffn_down_exps.weight: IQ1_M
999 blk.37.ffn_down_shexp.weight: Q5_K
1000 blk.37.ffn_gate_exps.weight: IQ2_XXS
1001 blk.37.ffn_gate_inp.weight: BF16
1002 blk.37.ffn_gate_shexp.weight: Q5_K
1003 blk.37.ffn_norm.weight: F32
1004 blk.37.ffn_up_exps.weight: IQ2_XXS
1005 blk.37.ffn_up_shexp.weight: Q5_K
1006 blk.37.hc_attn_base.weight: F32
1007 blk.37.hc_attn_fn.weight: F32
1008 blk.37.hc_attn_scale.weight: F32
1009 blk.37.hc_ffn_base.weight: F32
1010 blk.37.hc_ffn_fn.weight: F32
1011 blk.37.hc_ffn_scale.weight: F32
1012 blk.38.attn_kv.weight: Q8_0
1013 blk.38.attn_kv_a_norm.weight: F32
1014 blk.38.attn_norm.weight: F32
1015 blk.38.attn_output_a.weight: Q5_K
1016 blk.38.attn_output_b.weight: Q5_K
1017 blk.38.attn_q_a.weight: Q8_0
1018 blk.38.attn_q_a_norm.weight: F32
1019 blk.38.attn_q_b.weight: Q5_K
1020 blk.38.attn_sinks.weight: F32
1021 blk.38.exp_probs_b.bias: F32
1022 blk.38.exp_probs_b_vl.bias: F32
1023 blk.38.ffn_down_exps.weight: IQ2_XXS
1024 blk.38.ffn_down_shexp.weight: Q5_K
1025 blk.38.ffn_gate_exps.weight: IQ2_XS
1026 blk.38.ffn_gate_inp.weight: BF16
1027 blk.38.ffn_gate_shexp.weight: Q5_K
1028 blk.38.ffn_norm.weight: F32
1029 blk.38.ffn_up_exps.weight: IQ2_XS
1030 blk.38.ffn_up_shexp.weight: Q5_K
1031 blk.38.hc_attn_base.weight: F32
1032 blk.38.hc_attn_fn.weight: F32
1033 blk.38.hc_attn_scale.weight: F32
1034 blk.38.hc_ffn_base.weight: F32
1035 blk.38.hc_ffn_fn.weight: F32
1036 blk.38.hc_ffn_scale.weight: F32
1037 blk.39.attn_kv.weight: Q8_0
1038 blk.39.attn_kv_a_norm.weight: F32
1039 blk.39.attn_norm.weight: F32
1040 blk.39.attn_output_a.weight: Q5_K
1041 blk.39.attn_output_b.weight: Q5_K
1042 blk.39.attn_q_a.weight: Q8_0
1043 blk.39.attn_q_a_norm.weight: F32
1044 blk.39.attn_q_b.weight: Q5_K
1045 blk.39.attn_sinks.weight: F32
1046 blk.39.exp_probs_b.bias: F32
1047 blk.39.exp_probs_b_vl.bias: F32
1048 blk.39.ffn_down_exps.weight: IQ2_XS
1049 blk.39.ffn_down_shexp.weight: Q5_K
1050 blk.39.ffn_gate_exps.weight: IQ2_XS
1051 blk.39.ffn_gate_inp.weight: BF16
1052 blk.39.ffn_gate_shexp.weight: Q5_K
1053 blk.39.ffn_norm.weight: F32
1054 blk.39.ffn_up_exps.weight: IQ2_XS
1055 blk.39.ffn_up_shexp.weight: Q5_K
1056 blk.39.hc_attn_base.weight: F32
1057 blk.39.hc_attn_fn.weight: F32
1058 blk.39.hc_attn_scale.weight: F32
1059 blk.39.hc_ffn_base.weight: F32
1060 blk.39.hc_ffn_fn.weight: F32
1061 blk.39.hc_ffn_scale.weight: F32
1062