happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Fix merge conflicts, add more schemas

This commit is contained in:
Colin Kealty 2026-03-04 13:42:59 -05:00
parent 99119ceaf4
commit a3ff1940e9
13 changed files with 19689 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
tests/CMakeLists.txt
View File

@ -274,6 +274,12 @@ if (TARGET cpp-httplib)
add_executable(test-gguf-model-data test-gguf-model-data.cpp)
target_link_libraries(test-gguf-model-data PRIVATE gguf-model-data common)
llama_test(test-gguf-model-data LABEL "model")
# test-quant-type-selection requires gguf-model-data for remote model metadata
llama_build_and_test(test-quant-type-selection.cpp LABEL "model")
target_link_libraries(test-quant-type-selection PRIVATE gguf-model-data)
target_compile_definitions(test-quant-type-selection PRIVATE
SNAPSHOT_DIR="${CMAKE_CURRENT_SOURCE_DIR}/snapshots")
endif()
endif()

1
tests/gguf-model-data.cpp
View File

@ -151,6 +151,7 @@ static bool gguf_read_uint32_val(gguf_buf_reader & r, int32_t vtype, uint32_t &
return false;
}
}
return true;
}
if (vtype == GGUF_TYPE_UINT8) {
uint8_t v;

1956
tests/snapshots/deepseek-v3.1.schema Normal file
View File

File diff suppressed because it is too large Load Diff

1455
tests/snapshots/gemma-3-4b-it.schema Normal file
View File

File diff suppressed because it is too large Load Diff

1616
tests/snapshots/gpt-oss-120b.schema Normal file
View File

File diff suppressed because it is too large Load Diff

3899
tests/snapshots/meta-llama-3.1-70b-instruct.schema Normal file
View File

File diff suppressed because it is too large Load Diff

696
tests/snapshots/nemotron-nano-3-30b-a3b.schema Normal file
View File

@ -0,0 +1,696 @@
# Model: Nemotron-Nano-3-30B-A3B
# n_embd=2688, n_ff=0, n_vocab=131072, n_layer=52, n_head=32, n_head_kv=0, n_expert=128
[F32] f32
output.weight q6_K
[F16] f16
output.weight q6_K
[Q4_0] q4_0
output.weight q6_K
[Q4_1] q4_1
output.weight q6_K
[Q8_0] q8_0
[Q5_0] q5_0
output.weight q6_K
[Q5_1] q5_1
output.weight q6_K
[Q2_K] q2_K
output.weight q8_0
blk.1.ffn_down_exps.weight q3_K
blk.1.ffn_down_shexp.weight q3_K
blk.3.ffn_down_exps.weight q3_K
blk.3.ffn_down_shexp.weight q3_K
blk.5.attn_output.weight q3_K
blk.5.attn_v.weight q3_K
blk.6.ffn_down_exps.weight q3_K
blk.6.ffn_down_shexp.weight q3_K
blk.8.ffn_down_exps.weight q3_K
blk.8.ffn_down_shexp.weight q3_K
blk.10.ffn_down_exps.weight q3_K
blk.10.ffn_down_shexp.weight q3_K
blk.12.attn_output.weight q3_K
blk.12.attn_v.weight q3_K
blk.13.ffn_down_exps.weight q3_K
blk.13.ffn_down_shexp.weight q3_K
blk.15.ffn_down_exps.weight q3_K
blk.15.ffn_down_shexp.weight q3_K
blk.17.ffn_down_exps.weight q3_K
blk.17.ffn_down_shexp.weight q3_K
blk.19.attn_output.weight q3_K
blk.19.attn_v.weight q3_K
blk.20.ffn_down_exps.weight q3_K
blk.20.ffn_down_shexp.weight q3_K
blk.22.ffn_down_exps.weight q3_K
blk.22.ffn_down_shexp.weight q3_K
blk.24.ffn_down_exps.weight q3_K
blk.24.ffn_down_shexp.weight q3_K
blk.26.attn_output.weight q3_K
blk.26.attn_v.weight q3_K
blk.27.ffn_down_exps.weight q3_K
blk.27.ffn_down_shexp.weight q3_K
blk.29.ffn_down_exps.weight q3_K
blk.29.ffn_down_shexp.weight q3_K
blk.31.ffn_down_exps.weight q3_K
blk.31.ffn_down_shexp.weight q3_K
blk.33.attn_output.weight q3_K
blk.33.attn_v.weight q3_K
blk.34.ffn_down_exps.weight q3_K
blk.34.ffn_down_shexp.weight q3_K
blk.36.ffn_down_exps.weight q3_K
blk.36.ffn_down_shexp.weight q3_K
blk.38.ffn_down_exps.weight q3_K
blk.38.ffn_down_shexp.weight q3_K
blk.40.ffn_down_exps.weight q3_K
blk.40.ffn_down_shexp.weight q3_K
blk.42.attn_output.weight q3_K
blk.42.attn_v.weight q3_K
blk.43.ffn_down_exps.weight q3_K
blk.43.ffn_down_shexp.weight q3_K
blk.45.ffn_down_exps.weight q3_K
blk.45.ffn_down_shexp.weight q3_K
blk.47.ffn_down_exps.weight q3_K
blk.47.ffn_down_shexp.weight q3_K
blk.49.ffn_down_exps.weight q3_K
blk.49.ffn_down_shexp.weight q3_K
blk.51.ffn_down_exps.weight q3_K
blk.51.ffn_down_shexp.weight q3_K
[Q3_K_S] q3_K
output.weight q8_0
[Q3_K_M] q3_K
output.weight q8_0
blk.1.ffn_down_exps.weight q5_K
blk.1.ffn_down_shexp.weight q5_K
blk.3.ffn_down_exps.weight q4_K
blk.3.ffn_down_shexp.weight q4_K
blk.5.attn_output.weight q4_K
blk.5.attn_v.weight q5_K
blk.6.ffn_down_exps.weight q4_K
blk.6.ffn_down_shexp.weight q4_K
blk.8.ffn_down_exps.weight q4_K
blk.8.ffn_down_shexp.weight q4_K
blk.10.ffn_down_exps.weight q4_K
blk.10.ffn_down_shexp.weight q4_K
blk.12.attn_output.weight q4_K
blk.12.attn_v.weight q5_K
blk.13.ffn_down_exps.weight q4_K
blk.13.ffn_down_shexp.weight q4_K
blk.15.ffn_down_exps.weight q4_K
blk.15.ffn_down_shexp.weight q4_K
blk.17.ffn_down_exps.weight q4_K
blk.17.ffn_down_shexp.weight q4_K
blk.19.attn_output.weight q4_K
blk.19.attn_v.weight q4_K
blk.20.ffn_down_exps.weight q4_K
blk.20.ffn_down_shexp.weight q4_K
blk.22.ffn_down_exps.weight q4_K
blk.22.ffn_down_shexp.weight q4_K
blk.24.ffn_down_exps.weight q4_K
blk.24.ffn_down_shexp.weight q4_K
blk.26.attn_output.weight q4_K
blk.26.attn_v.weight q4_K
blk.27.ffn_down_exps.weight q4_K
blk.27.ffn_down_shexp.weight q4_K
blk.29.ffn_down_exps.weight q4_K
blk.29.ffn_down_shexp.weight q4_K
blk.31.ffn_down_exps.weight q4_K
blk.31.ffn_down_shexp.weight q4_K
blk.33.attn_output.weight q4_K
blk.33.attn_v.weight q4_K
blk.34.ffn_down_exps.weight q4_K
blk.34.ffn_down_shexp.weight q4_K
blk.36.ffn_down_exps.weight q4_K
blk.36.ffn_down_shexp.weight q4_K
blk.38.ffn_down_exps.weight q4_K
blk.38.ffn_down_shexp.weight q4_K
blk.40.ffn_down_exps.weight q4_K
blk.40.ffn_down_shexp.weight q4_K
blk.42.attn_output.weight q4_K
blk.42.attn_v.weight q4_K
blk.43.ffn_down_exps.weight q4_K
blk.43.ffn_down_shexp.weight q4_K
blk.45.ffn_down_exps.weight q4_K
blk.45.ffn_down_shexp.weight q4_K
blk.47.ffn_down_exps.weight q4_K
blk.47.ffn_down_shexp.weight q4_K
blk.49.ffn_down_exps.weight q4_K
blk.49.ffn_down_shexp.weight q4_K
blk.51.ffn_down_exps.weight q4_K
blk.51.ffn_down_shexp.weight q4_K
[Q3_K_L] q3_K
output.weight q8_0
blk.1.ffn_down_exps.weight q5_K
blk.1.ffn_down_shexp.weight q5_K
blk.3.ffn_down_exps.weight q5_K
blk.3.ffn_down_shexp.weight q5_K
blk.5.attn_output.weight q5_K
blk.5.attn_v.weight q5_K
blk.6.ffn_down_exps.weight q5_K
blk.6.ffn_down_shexp.weight q5_K
blk.8.ffn_down_exps.weight q5_K
blk.8.ffn_down_shexp.weight q5_K
blk.10.ffn_down_exps.weight q5_K
blk.10.ffn_down_shexp.weight q5_K
blk.12.attn_output.weight q5_K
blk.12.attn_v.weight q5_K
blk.13.ffn_down_exps.weight q5_K
blk.13.ffn_down_shexp.weight q5_K
blk.15.ffn_down_exps.weight q5_K
blk.15.ffn_down_shexp.weight q5_K
blk.17.ffn_down_exps.weight q5_K
blk.17.ffn_down_shexp.weight q5_K
blk.19.attn_output.weight q5_K
blk.19.attn_v.weight q5_K
blk.20.ffn_down_exps.weight q5_K
blk.20.ffn_down_shexp.weight q5_K
blk.22.ffn_down_exps.weight q5_K
blk.22.ffn_down_shexp.weight q5_K
blk.24.ffn_down_exps.weight q5_K
blk.24.ffn_down_shexp.weight q5_K
blk.26.attn_output.weight q5_K
blk.26.attn_v.weight q5_K
blk.27.ffn_down_exps.weight q5_K
blk.27.ffn_down_shexp.weight q5_K
blk.29.ffn_down_exps.weight q5_K
blk.29.ffn_down_shexp.weight q5_K
blk.31.ffn_down_exps.weight q5_K
blk.31.ffn_down_shexp.weight q5_K
blk.33.attn_output.weight q5_K
blk.33.attn_v.weight q5_K
blk.34.ffn_down_exps.weight q5_K
blk.34.ffn_down_shexp.weight q5_K
blk.36.ffn_down_exps.weight q5_K
blk.36.ffn_down_shexp.weight q5_K
blk.38.ffn_down_exps.weight q5_K
blk.38.ffn_down_shexp.weight q5_K
blk.40.ffn_down_exps.weight q5_K
blk.40.ffn_down_shexp.weight q5_K
blk.42.attn_output.weight q5_K
blk.42.attn_v.weight q5_K
blk.43.ffn_down_exps.weight q5_K
blk.43.ffn_down_shexp.weight q5_K
blk.45.ffn_down_exps.weight q5_K
blk.45.ffn_down_shexp.weight q5_K
blk.47.ffn_down_exps.weight q5_K
blk.47.ffn_down_shexp.weight q5_K
blk.49.ffn_down_exps.weight q5_K
blk.49.ffn_down_shexp.weight q5_K
blk.51.ffn_down_exps.weight q5_K
blk.51.ffn_down_shexp.weight q5_K
[Q4_K_S] q4_K
output.weight q8_0
blk.1.ffn_down_exps.weight q5_K
blk.1.ffn_down_shexp.weight q5_K
blk.3.ffn_down_exps.weight q5_K
blk.3.ffn_down_shexp.weight q5_K
blk.5.attn_v.weight q5_K
blk.12.attn_v.weight q5_K
blk.19.attn_v.weight q5_K
blk.26.attn_v.weight q5_K
[Q4_K_M] q4_K
output.weight q8_0
blk.1.ffn_down_exps.weight q6_K
blk.1.ffn_down_shexp.weight q6_K
blk.3.ffn_down_exps.weight q6_K
blk.3.ffn_down_shexp.weight q6_K
blk.8.ffn_down_exps.weight q6_K
blk.8.ffn_down_shexp.weight q6_K
blk.17.ffn_down_exps.weight q6_K
blk.17.ffn_down_shexp.weight q6_K
blk.19.attn_v.weight q6_K
blk.20.ffn_down_exps.weight q6_K
blk.20.ffn_down_shexp.weight q6_K
blk.29.ffn_down_exps.weight q6_K
blk.29.ffn_down_shexp.weight q6_K
blk.38.ffn_down_exps.weight q6_K
blk.38.ffn_down_shexp.weight q6_K
blk.42.attn_v.weight q6_K
blk.45.ffn_down_exps.weight q6_K
blk.45.ffn_down_shexp.weight q6_K
blk.47.ffn_down_exps.weight q6_K
blk.47.ffn_down_shexp.weight q6_K
blk.49.ffn_down_exps.weight q6_K
blk.49.ffn_down_shexp.weight q6_K
blk.51.ffn_down_exps.weight q6_K
blk.51.ffn_down_shexp.weight q6_K
[Q5_K_S] q5_K
output.weight q8_0
[Q5_K_M] q5_K
output.weight q8_0
blk.1.ffn_down_exps.weight q6_K
blk.1.ffn_down_shexp.weight q6_K
blk.3.ffn_down_exps.weight q6_K
blk.3.ffn_down_shexp.weight q6_K
blk.8.ffn_down_exps.weight q6_K
blk.8.ffn_down_shexp.weight q6_K
blk.17.ffn_down_exps.weight q6_K
blk.17.ffn_down_shexp.weight q6_K
blk.19.attn_v.weight q6_K
blk.20.ffn_down_exps.weight q6_K
blk.20.ffn_down_shexp.weight q6_K
blk.29.ffn_down_exps.weight q6_K
blk.29.ffn_down_shexp.weight q6_K
blk.38.ffn_down_exps.weight q6_K
blk.38.ffn_down_shexp.weight q6_K
blk.42.attn_v.weight q6_K
blk.45.ffn_down_exps.weight q6_K
blk.45.ffn_down_shexp.weight q6_K
blk.47.ffn_down_exps.weight q6_K
blk.47.ffn_down_shexp.weight q6_K
blk.49.ffn_down_exps.weight q6_K
blk.49.ffn_down_shexp.weight q6_K
blk.51.ffn_down_exps.weight q6_K
blk.51.ffn_down_shexp.weight q6_K
[Q6_K] q6_K
output.weight q8_0
[IQ2_XXS] iq2_xxs
output.weight q8_0
token_embd.weight q2_K
blk.1.ffn_down_exps.weight q2_K
blk.1.ffn_down_shexp.weight q2_K
blk.3.ffn_down_exps.weight q2_K
blk.3.ffn_down_shexp.weight q2_K
blk.5.attn_v.weight q4_K
blk.6.ffn_down_exps.weight q2_K
blk.6.ffn_down_shexp.weight q2_K
blk.12.attn_v.weight q4_K
blk.19.attn_v.weight q4_K
blk.26.attn_v.weight q4_K
blk.33.attn_v.weight q4_K
blk.42.attn_v.weight q4_K
[IQ2_XS] iq2_xs
output.weight q8_0
token_embd.weight q2_K
blk.1.ffn_down_exps.weight q2_K
blk.1.ffn_down_shexp.weight q2_K
blk.3.ffn_down_exps.weight q2_K
blk.3.ffn_down_shexp.weight q2_K
blk.5.attn_v.weight q4_K
blk.6.ffn_down_exps.weight q2_K
blk.6.ffn_down_shexp.weight q2_K
blk.12.attn_v.weight q4_K
blk.19.attn_v.weight q4_K
blk.26.attn_v.weight q4_K
blk.33.attn_v.weight q4_K
blk.42.attn_v.weight q4_K
[Q2_K_S] q2_K
output.weight q8_0
blk.1.ffn_down_exps.weight q4_K
blk.1.ffn_down_shexp.weight q4_K
blk.3.ffn_down_exps.weight q4_K
blk.3.ffn_down_shexp.weight q4_K
[IQ3_XS] iq3_s
output.weight q8_0
blk.5.attn_k.weight iq3_xxs
blk.5.attn_q.weight iq3_xxs
blk.6.ffn_up_exps.weight iq3_xxs
blk.6.ffn_up_shexp.weight iq3_xxs
blk.8.ffn_up_exps.weight iq3_xxs
blk.8.ffn_up_shexp.weight iq3_xxs
blk.10.ffn_up_exps.weight iq3_xxs
blk.10.ffn_up_shexp.weight iq3_xxs
blk.12.attn_k.weight iq3_xxs
blk.12.attn_q.weight iq3_xxs
blk.13.ffn_up_exps.weight iq3_xxs
blk.13.ffn_up_shexp.weight iq3_xxs
blk.15.ffn_up_exps.weight iq3_xxs
blk.15.ffn_up_shexp.weight iq3_xxs
blk.17.ffn_up_exps.weight iq3_xxs
blk.17.ffn_up_shexp.weight iq3_xxs
blk.19.attn_k.weight iq3_xxs
blk.19.attn_q.weight iq3_xxs
blk.20.ffn_up_exps.weight iq3_xxs
blk.20.ffn_up_shexp.weight iq3_xxs
blk.22.ffn_up_exps.weight iq3_xxs
blk.22.ffn_up_shexp.weight iq3_xxs
blk.24.ffn_up_exps.weight iq3_xxs
blk.24.ffn_up_shexp.weight iq3_xxs
blk.26.attn_k.weight iq3_xxs
blk.26.attn_q.weight iq3_xxs
blk.27.ffn_up_exps.weight iq3_xxs
blk.27.ffn_up_shexp.weight iq3_xxs
blk.29.ffn_up_exps.weight iq3_xxs
blk.29.ffn_up_shexp.weight iq3_xxs
blk.31.ffn_up_exps.weight iq3_xxs
blk.31.ffn_up_shexp.weight iq3_xxs
blk.33.attn_k.weight iq3_xxs
blk.33.attn_q.weight iq3_xxs
blk.34.ffn_up_exps.weight iq3_xxs
blk.34.ffn_up_shexp.weight iq3_xxs
blk.36.ffn_up_exps.weight iq3_xxs
blk.36.ffn_up_shexp.weight iq3_xxs
blk.38.ffn_up_exps.weight iq3_xxs
blk.38.ffn_up_shexp.weight iq3_xxs
blk.40.ffn_up_exps.weight iq3_xxs
blk.40.ffn_up_shexp.weight iq3_xxs
blk.42.attn_k.weight iq3_xxs
blk.42.attn_q.weight iq3_xxs
blk.43.ffn_up_exps.weight iq3_xxs
blk.43.ffn_up_shexp.weight iq3_xxs
[IQ3_XXS] iq3_xxs
output.weight q8_0
token_embd.weight iq3_s
blk.1.ffn_down_exps.weight q4_K
blk.1.ffn_down_shexp.weight q4_K
blk.3.ffn_down_exps.weight q4_K
blk.3.ffn_down_shexp.weight q4_K
blk.5.attn_k.weight iq2_s
blk.5.attn_output.weight iq3_s
blk.5.attn_q.weight iq2_s
blk.5.attn_v.weight iq3_s
blk.6.ffn_down_exps.weight q3_K
blk.6.ffn_down_shexp.weight q3_K
blk.8.ffn_down_exps.weight q3_K
blk.8.ffn_down_shexp.weight q3_K
blk.10.ffn_down_exps.weight q3_K
blk.10.ffn_down_shexp.weight q3_K
blk.12.attn_k.weight iq2_s
blk.12.attn_output.weight iq3_s
blk.12.attn_q.weight iq2_s
blk.12.attn_v.weight iq3_s
blk.13.ffn_down_exps.weight q3_K
blk.13.ffn_down_shexp.weight q3_K
blk.15.ffn_down_exps.weight q3_K
blk.15.ffn_down_shexp.weight q3_K
blk.17.ffn_down_exps.weight q3_K
blk.17.ffn_down_shexp.weight q3_K
blk.19.attn_k.weight iq2_s
blk.19.attn_output.weight iq3_s
blk.19.attn_q.weight iq2_s
blk.19.attn_v.weight iq3_s
blk.20.ffn_down_exps.weight q3_K
blk.20.ffn_down_shexp.weight q3_K
blk.22.ffn_down_exps.weight q3_K
blk.22.ffn_down_shexp.weight q3_K
blk.24.ffn_down_exps.weight q3_K
blk.24.ffn_down_shexp.weight q3_K
blk.26.attn_k.weight iq2_s
blk.26.attn_output.weight iq3_s
blk.26.attn_q.weight iq2_s
blk.26.attn_v.weight iq3_s
blk.27.ffn_down_exps.weight q3_K
blk.27.ffn_down_shexp.weight q3_K
blk.29.ffn_down_exps.weight q3_K
blk.29.ffn_down_shexp.weight q3_K
blk.31.ffn_down_exps.weight q3_K
blk.31.ffn_down_shexp.weight q3_K
blk.33.attn_k.weight iq2_s
blk.33.attn_output.weight iq3_s
blk.33.attn_q.weight iq2_s
blk.33.attn_v.weight iq3_s
blk.34.ffn_down_exps.weight q3_K
blk.34.ffn_down_shexp.weight q3_K
blk.36.ffn_down_exps.weight q3_K
blk.36.ffn_down_shexp.weight q3_K
blk.38.ffn_down_exps.weight q3_K
blk.38.ffn_down_shexp.weight q3_K
blk.40.ffn_down_exps.weight q3_K
blk.40.ffn_down_shexp.weight q3_K
blk.42.attn_k.weight iq2_s
blk.42.attn_output.weight iq3_s
blk.42.attn_q.weight iq2_s
blk.42.attn_v.weight iq3_s
blk.43.ffn_down_exps.weight q3_K
blk.43.ffn_down_shexp.weight q3_K
blk.45.ffn_down_exps.weight q3_K
blk.45.ffn_down_shexp.weight q3_K
blk.47.ffn_down_exps.weight q3_K
blk.47.ffn_down_shexp.weight q3_K
blk.49.ffn_down_exps.weight q3_K
blk.49.ffn_down_shexp.weight q3_K
blk.51.ffn_down_exps.weight q3_K
blk.51.ffn_down_shexp.weight q3_K
[IQ1_S] iq1_s
output.weight q8_0
token_embd.weight q2_K
blk.1.ffn_down_exps.weight q2_K
blk.1.ffn_down_shexp.weight q2_K
blk.3.ffn_down_exps.weight q2_K
blk.3.ffn_down_shexp.weight q2_K
blk.5.attn_output.weight iq2_xxs
blk.5.attn_v.weight q4_K
blk.6.ffn_down_exps.weight q2_K
blk.6.ffn_down_shexp.weight q2_K
blk.12.attn_output.weight iq2_xxs
blk.12.attn_v.weight q4_K
blk.19.attn_output.weight iq2_xxs
blk.19.attn_v.weight q4_K
blk.26.attn_output.weight iq2_xxs
blk.26.attn_v.weight q4_K
blk.33.attn_output.weight iq2_xxs
blk.33.attn_v.weight q4_K
blk.42.attn_output.weight iq2_xxs
blk.42.attn_v.weight q4_K
[IQ4_NL] iq4_nl
output.weight q6_K
blk.1.ffn_down_exps.weight q5_K
blk.1.ffn_down_shexp.weight q5_K
blk.3.ffn_down_exps.weight q5_K
blk.3.ffn_down_shexp.weight q5_K
[IQ3_S] iq3_s
output.weight q8_0
[IQ3_M] iq3_s
output.weight q8_0
blk.1.ffn_down_exps.weight q4_K
blk.1.ffn_down_shexp.weight q4_K
blk.3.ffn_down_exps.weight q4_K
blk.3.ffn_down_shexp.weight q4_K
blk.5.attn_output.weight q4_K
blk.5.attn_v.weight q4_K
blk.12.attn_output.weight q4_K
blk.12.attn_v.weight q4_K
blk.19.attn_output.weight q4_K
blk.19.attn_v.weight q4_K
blk.26.attn_output.weight q4_K
blk.26.attn_v.weight q4_K
blk.33.attn_output.weight q4_K
blk.33.attn_v.weight q4_K
blk.42.attn_output.weight q4_K
blk.42.attn_v.weight q4_K
[IQ2_S] iq2_xs
output.weight q8_0
token_embd.weight iq3_s
blk.1.ffn_down_exps.weight iq3_s
blk.1.ffn_down_shexp.weight iq3_s
blk.3.ffn_down_exps.weight iq3_s
blk.3.ffn_down_shexp.weight iq3_s
blk.5.attn_output.weight iq3_s
blk.5.attn_v.weight q4_K
blk.6.ffn_down_exps.weight iq3_s
blk.6.ffn_down_shexp.weight iq3_s
blk.12.attn_output.weight iq3_s
blk.12.attn_v.weight q4_K
blk.19.attn_output.weight iq3_s
blk.19.attn_v.weight q4_K
blk.26.attn_output.weight iq3_s
blk.26.attn_v.weight q4_K
blk.33.attn_output.weight iq3_s
blk.33.attn_v.weight q4_K
blk.42.attn_output.weight iq3_s
blk.42.attn_v.weight q4_K
[IQ2_M] iq2_s
output.weight q8_0
token_embd.weight iq3_s
blk.1.ffn_down_exps.weight iq3_s
blk.1.ffn_down_shexp.weight iq3_s
blk.3.ffn_down_exps.weight iq3_s
blk.3.ffn_down_shexp.weight iq3_s
blk.5.attn_output.weight iq3_s
blk.5.attn_v.weight q4_K
blk.6.ffn_down_exps.weight iq3_s
blk.6.ffn_down_shexp.weight iq3_s
blk.12.attn_output.weight iq3_s
blk.12.attn_v.weight q4_K
blk.19.attn_output.weight iq3_s
blk.19.attn_v.weight q4_K
blk.26.attn_output.weight iq3_s
blk.26.attn_v.weight q4_K
blk.33.attn_output.weight iq3_s
blk.33.attn_v.weight q4_K
blk.42.attn_output.weight iq3_s
blk.42.attn_v.weight q4_K
[IQ4_XS] iq4_xs
output.weight q8_0
blk.1.ffn_down_exps.weight q5_K
blk.1.ffn_down_shexp.weight q5_K
blk.3.ffn_down_exps.weight q5_K
blk.3.ffn_down_shexp.weight q5_K
[IQ1_M] iq1_m
output.weight q8_0
token_embd.weight q2_K
blk.1.ffn_down_exps.weight q2_K
blk.1.ffn_down_shexp.weight q2_K
blk.3.ffn_down_exps.weight q2_K
blk.3.ffn_down_shexp.weight q2_K
blk.5.attn_output.weight iq2_xxs
blk.5.attn_v.weight q4_K
blk.6.ffn_down_exps.weight q2_K
blk.6.ffn_down_shexp.weight q2_K
blk.12.attn_output.weight iq2_xxs
blk.12.attn_v.weight q4_K
blk.19.attn_output.weight iq2_xxs
blk.19.attn_v.weight q4_K
blk.26.attn_output.weight iq2_xxs
blk.26.attn_v.weight q4_K
blk.33.attn_output.weight iq2_xxs
blk.33.attn_v.weight q4_K
blk.42.attn_output.weight iq2_xxs
blk.42.attn_v.weight q4_K
[BF16] bf16
output.weight q6_K
[TQ1_0] tq1_0
output.weight q8_0
token_embd.weight q4_K
[TQ2_0] tq2_0
output.weight q8_0
token_embd.weight q4_K
[MXFP4_MOE] mxfp4
output.weight q8_0
token_embd.weight q8_0
blk.0.ssm_in.weight q8_0
blk.0.ssm_out.weight q8_0
blk.1.ffn_down_shexp.weight q8_0
blk.1.ffn_up_shexp.weight q8_0
blk.2.ssm_in.weight q8_0
blk.2.ssm_out.weight q8_0
blk.3.ffn_down_shexp.weight q8_0
blk.3.ffn_up_shexp.weight q8_0
blk.4.ssm_in.weight q8_0
blk.4.ssm_out.weight q8_0
blk.5.attn_k.weight q8_0
blk.5.attn_output.weight q8_0
blk.5.attn_q.weight q8_0
blk.5.attn_v.weight q8_0
blk.6.ffn_down_shexp.weight q8_0
blk.6.ffn_up_shexp.weight q8_0
blk.7.ssm_in.weight q8_0
blk.7.ssm_out.weight q8_0
blk.8.ffn_down_shexp.weight q8_0
blk.8.ffn_up_shexp.weight q8_0
blk.9.ssm_in.weight q8_0
blk.9.ssm_out.weight q8_0
blk.10.ffn_down_shexp.weight q8_0
blk.10.ffn_up_shexp.weight q8_0
blk.11.ssm_in.weight q8_0
blk.11.ssm_out.weight q8_0
blk.12.attn_k.weight q8_0
blk.12.attn_output.weight q8_0
blk.12.attn_q.weight q8_0
blk.12.attn_v.weight q8_0
blk.13.ffn_down_shexp.weight q8_0
blk.13.ffn_up_shexp.weight q8_0
blk.14.ssm_in.weight q8_0
blk.14.ssm_out.weight q8_0
blk.15.ffn_down_shexp.weight q8_0
blk.15.ffn_up_shexp.weight q8_0
blk.16.ssm_in.weight q8_0
blk.16.ssm_out.weight q8_0
blk.17.ffn_down_shexp.weight q8_0
blk.17.ffn_up_shexp.weight q8_0
blk.18.ssm_in.weight q8_0
blk.18.ssm_out.weight q8_0
blk.19.attn_k.weight q8_0
blk.19.attn_output.weight q8_0
blk.19.attn_q.weight q8_0
blk.19.attn_v.weight q8_0
blk.20.ffn_down_shexp.weight q8_0
blk.20.ffn_up_shexp.weight q8_0
blk.21.ssm_in.weight q8_0
blk.21.ssm_out.weight q8_0
blk.22.ffn_down_shexp.weight q8_0
blk.22.ffn_up_shexp.weight q8_0
blk.23.ssm_in.weight q8_0
blk.23.ssm_out.weight q8_0
blk.24.ffn_down_shexp.weight q8_0
blk.24.ffn_up_shexp.weight q8_0
blk.25.ssm_in.weight q8_0
blk.25.ssm_out.weight q8_0
blk.26.attn_k.weight q8_0
blk.26.attn_output.weight q8_0
blk.26.attn_q.weight q8_0
blk.26.attn_v.weight q8_0
blk.27.ffn_down_shexp.weight q8_0
blk.27.ffn_up_shexp.weight q8_0
blk.28.ssm_in.weight q8_0
blk.28.ssm_out.weight q8_0
blk.29.ffn_down_shexp.weight q8_0
blk.29.ffn_up_shexp.weight q8_0
blk.30.ssm_in.weight q8_0
blk.30.ssm_out.weight q8_0
blk.31.ffn_down_shexp.weight q8_0
blk.31.ffn_up_shexp.weight q8_0
blk.32.ssm_in.weight q8_0
blk.32.ssm_out.weight q8_0
blk.33.attn_k.weight q8_0
blk.33.attn_output.weight q8_0
blk.33.attn_q.weight q8_0
blk.33.attn_v.weight q8_0
blk.34.ffn_down_shexp.weight q8_0
blk.34.ffn_up_shexp.weight q8_0
blk.35.ssm_in.weight q8_0
blk.35.ssm_out.weight q8_0
blk.36.ffn_down_shexp.weight q8_0
blk.36.ffn_up_shexp.weight q8_0
blk.37.ssm_in.weight q8_0
blk.37.ssm_out.weight q8_0
blk.38.ffn_down_shexp.weight q8_0
blk.38.ffn_up_shexp.weight q8_0
blk.39.ssm_in.weight q8_0
blk.39.ssm_out.weight q8_0
blk.40.ffn_down_shexp.weight q8_0
blk.40.ffn_up_shexp.weight q8_0
blk.41.ssm_in.weight q8_0
blk.41.ssm_out.weight q8_0
blk.42.attn_k.weight q8_0
blk.42.attn_output.weight q8_0
blk.42.attn_q.weight q8_0
blk.42.attn_v.weight q8_0
blk.43.ffn_down_shexp.weight q8_0
blk.43.ffn_up_shexp.weight q8_0
blk.44.ssm_in.weight q8_0
blk.44.ssm_out.weight q8_0
blk.45.ffn_down_shexp.weight q8_0
blk.45.ffn_up_shexp.weight q8_0
blk.46.ssm_in.weight q8_0
blk.46.ssm_out.weight q8_0
blk.47.ffn_down_shexp.weight q8_0
blk.47.ffn_up_shexp.weight q8_0
blk.48.ssm_in.weight q8_0
blk.48.ssm_out.weight q8_0
blk.49.ffn_down_shexp.weight q8_0
blk.49.ffn_up_shexp.weight q8_0
blk.50.ssm_in.weight q8_0
blk.50.ssm_out.weight q8_0
blk.51.ffn_down_shexp.weight q8_0
blk.51.ffn_up_shexp.weight q8_0

1908
tests/snapshots/qwen3-14b.schema Normal file
View File

File diff suppressed because it is too large Load Diff

1713
tests/snapshots/qwen3-coder-next.schema Normal file
View File

File diff suppressed because it is too large Load Diff

1837
tests/snapshots/qwen3.5-27b.schema Normal file
View File

File diff suppressed because it is too large Load Diff

2148
tests/snapshots/qwen3.5-397b-a17b.schema Normal file
View File

File diff suppressed because it is too large Load Diff

2453
tests/snapshots/step-3.5-flash.schema Normal file
View File

File diff suppressed because it is too large Load Diff

2
tests/test-quant-type-selection.cpp
View File

@ -190,7 +190,7 @@ static const remote_model_spec model_specs[] = {
{ "ggml-org/gpt-oss-120b-GGUF", "mxfp4" },
{ "ggml-org/gemma-3-4b-it-GGUF", "Q8_0" },
{ "bartowski/Meta-Llama-3.1-70B-Instruct-GGUF", "Q4_K_M" },
{ "bartowski/deepseek-ai_DeepSeek-V3.1-GGUF", "IQ1_S" },
{ "bartowski/deepseek-ai_DeepSeek-V3.1-GGUF", "IQ1_M" },
{ "bartowski/Qwen_Qwen3.5-397B-A17B-GGUF", "IQ1_S" }, // TODO: swap with ggml-org if/when it's released
{ "bartowski/Qwen_Qwen3.5-27B-GGUF", "Q8_0" }, // TODO: swap with ggml-org if/when it's released
};