Skip to content

Commit 32bf33f

Browse files
localai-org-maint-botlocalai-org-maint-bot
authored
feat(gallery): add Genesis Hermes V7 variants (#11435)
Add the recommended APEX, compact, MTP-enabled, and Q8_K_P builds of Qwen3.6 35B A3B Genesis Hermes V7 with its shared multimodal projector. Assisted-by: Codex:gpt-5 Co-authored-by: localai-org-maint-bot <306269227+localai-org-maint-bot@users.noreply.github.com>
1 parent 7b9167e commit 32bf33f

1 file changed

Lines changed: 229 additions & 0 deletions

File tree

‎gallery/index.yaml‎

Lines changed: 229 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -928,6 +928,235 @@
928928
- filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V6-Q8_0/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
929929
sha256: c8e702344a81f8c226a914aa980ed6e1f604bce9374f1fed8e65c896908af414
930930
uri: https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V6-GGUF/resolve/main/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
931+
- &qwen3-6-35b-a3b-genesis-hermes-v7
932+
name: "qwen3.6-35b-a3b-genesis-hermes-v7"
933+
url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
934+
variants:
935+
- model: qwen3.6-35b-a3b-genesis-hermes-v7-apex-compact
936+
- model: qwen3.6-35b-a3b-genesis-hermes-v7-mtp-apex
937+
- model: qwen3.6-35b-a3b-genesis-hermes-v7-mtp-apex-compact
938+
- model: qwen3.6-35b-a3b-genesis-hermes-v7-q8-k-p
939+
urls:
940+
- https://huggingface.co/HauhauCS/Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive
941+
- https://huggingface.co/LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF
942+
description: |
943+
Qwen3.6-35B-A3B Genesis Hermes V7 is LuffyTheFox's Apache-2.0 multimodal,
944+
agentic derivative of HauhauCS's uncensored Qwen3.6-35B-A3B model. It
945+
combines Genesis tensor calibration with Hermes function-calling data while
946+
retaining the 35B mixture-of-experts architecture, roughly 3B active
947+
parameters per token, and the native 262K-token context window.
948+
949+
This entry's own payload uses the model card's recommended APEX GGUF and the
950+
shared F16 multimodal projector. Automatic variant selection may instead
951+
choose Compact APEX, an MTP-enabled APEX build, or Q8_K_P based on serving
952+
features and available memory. The model card recommends Jinja chat
953+
templates and at least a 128K context for its thinking behavior.
954+
license: "apache-2.0"
955+
tags:
956+
- llm
957+
- gguf
958+
- vision
959+
- multimodal
960+
- reasoning
961+
- coding
962+
- uncensored
963+
- moe
964+
last_checked: "2026-08-10"
965+
overrides:
966+
backend: llama-cpp
967+
function:
968+
automatic_tool_parsing_fallback: true
969+
grammar:
970+
disable: true
971+
known_usecases:
972+
- chat
973+
- vision
974+
mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
975+
options:
976+
- use_jinja:true
977+
parameters:
978+
min_p: 0
979+
model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX.gguf
980+
repeat_penalty: 1
981+
temperature: 0.6
982+
top_k: 20
983+
top_p: 0.95
984+
template:
985+
use_tokenizer_template: true
986+
files:
987+
- filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX.gguf
988+
sha256: 995c2a82e1aaca7947075891ead093e38d1e7dc5c0b54e4debd1d01058229ff9
989+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX.gguf
990+
- filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
991+
sha256: c8e702344a81f8c226a914aa980ed6e1f604bce9374f1fed8e65c896908af414
992+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
993+
- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-v7
994+
name: "qwen3.6-35b-a3b-genesis-hermes-v7-apex-compact"
995+
variants: []
996+
description: |
997+
Qwen3.6-35B-A3B Genesis Hermes V7 in the smaller APEX Compact GGUF format,
998+
with the shared F16 multimodal projector. This build preserves the model's
999+
multimodal, reasoning, coding, and agentic capabilities for hosts with less
1000+
memory than the recommended full APEX build.
1001+
overrides:
1002+
backend: llama-cpp
1003+
function:
1004+
automatic_tool_parsing_fallback: true
1005+
grammar:
1006+
disable: true
1007+
known_usecases:
1008+
- chat
1009+
- vision
1010+
mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1011+
options:
1012+
- use_jinja:true
1013+
parameters:
1014+
min_p: 0
1015+
model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX-Compact.gguf
1016+
repeat_penalty: 1
1017+
temperature: 0.6
1018+
top_k: 20
1019+
top_p: 0.95
1020+
template:
1021+
use_tokenizer_template: true
1022+
files:
1023+
- filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX-Compact.gguf
1024+
sha256: 41826ae69333ab4f3577d53c0a8b427b306c63c88c6200b4eada32dda28d26b9
1025+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-APEX-Compact.gguf
1026+
- filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1027+
sha256: c8e702344a81f8c226a914aa980ed6e1f604bce9374f1fed8e65c896908af414
1028+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1029+
- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-v7
1030+
name: "qwen3.6-35b-a3b-genesis-hermes-v7-mtp-apex"
1031+
variants: []
1032+
description: |
1033+
Qwen3.6-35B-A3B Genesis Hermes V7 in the full APEX GGUF format with native
1034+
multi-token prediction enabled for speculative decoding, plus the shared
1035+
F16 multimodal projector.
1036+
tags:
1037+
- llm
1038+
- gguf
1039+
- vision
1040+
- multimodal
1041+
- reasoning
1042+
- coding
1043+
- uncensored
1044+
- moe
1045+
- mtp
1046+
overrides:
1047+
backend: llama-cpp
1048+
function:
1049+
automatic_tool_parsing_fallback: true
1050+
grammar:
1051+
disable: true
1052+
known_usecases:
1053+
- chat
1054+
- vision
1055+
mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1056+
options:
1057+
- use_jinja:true
1058+
- spec_type:draft-mtp
1059+
- spec_n_max:6
1060+
- spec_p_min:0.75
1061+
parameters:
1062+
min_p: 0
1063+
model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX.gguf
1064+
repeat_penalty: 1
1065+
temperature: 0.6
1066+
top_k: 20
1067+
top_p: 0.95
1068+
template:
1069+
use_tokenizer_template: true
1070+
files:
1071+
- filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX.gguf
1072+
sha256: 63a3cd730a4f1e7f5b909c89550de32e5a27b33c697c62aed9da60a7bd4643ae
1073+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX.gguf
1074+
- filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1075+
sha256: c8e702344a81f8c226a914aa980ed6e1f604bce9374f1fed8e65c896908af414
1076+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1077+
- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-v7
1078+
name: "qwen3.6-35b-a3b-genesis-hermes-v7-mtp-apex-compact"
1079+
variants: []
1080+
description: |
1081+
Qwen3.6-35B-A3B Genesis Hermes V7 in the smaller APEX Compact GGUF format
1082+
with native multi-token prediction enabled for speculative decoding, plus
1083+
the shared F16 multimodal projector.
1084+
tags:
1085+
- llm
1086+
- gguf
1087+
- vision
1088+
- multimodal
1089+
- reasoning
1090+
- coding
1091+
- uncensored
1092+
- moe
1093+
- mtp
1094+
overrides:
1095+
backend: llama-cpp
1096+
function:
1097+
automatic_tool_parsing_fallback: true
1098+
grammar:
1099+
disable: true
1100+
known_usecases:
1101+
- chat
1102+
- vision
1103+
mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1104+
options:
1105+
- use_jinja:true
1106+
- spec_type:draft-mtp
1107+
- spec_n_max:6
1108+
- spec_p_min:0.75
1109+
parameters:
1110+
min_p: 0
1111+
model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX-Compact.gguf
1112+
repeat_penalty: 1
1113+
temperature: 0.6
1114+
top_k: 20
1115+
top_p: 0.95
1116+
template:
1117+
use_tokenizer_template: true
1118+
files:
1119+
- filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX-Compact/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX-Compact.gguf
1120+
sha256: 11edd0da4f28f587c0eb72ae2e417dfb0d02733360303045e21c9ba3aae8d114
1121+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-MTP-APEX-Compact.gguf
1122+
- filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1123+
sha256: c8e702344a81f8c226a914aa980ed6e1f604bce9374f1fed8e65c896908af414
1124+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1125+
- !!merge <<: *qwen3-6-35b-a3b-genesis-hermes-v7
1126+
name: "qwen3.6-35b-a3b-genesis-hermes-v7-q8-k-p"
1127+
variants: []
1128+
description: |
1129+
Qwen3.6-35B-A3B Genesis Hermes V7 in the high-quality Q8_K_P GGUF format,
1130+
with the shared F16 multimodal projector. This is the largest non-MTP build
1131+
in the published V7 set.
1132+
overrides:
1133+
backend: llama-cpp
1134+
function:
1135+
automatic_tool_parsing_fallback: true
1136+
grammar:
1137+
disable: true
1138+
known_usecases:
1139+
- chat
1140+
- vision
1141+
mmproj: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1142+
options:
1143+
- use_jinja:true
1144+
parameters:
1145+
min_p: 0
1146+
model: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-Q8_K_P/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-Q8_K_P.gguf
1147+
repeat_penalty: 1
1148+
temperature: 0.6
1149+
top_k: 20
1150+
top_p: 0.95
1151+
template:
1152+
use_tokenizer_template: true
1153+
files:
1154+
- filename: llama-cpp/models/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-Q8_K_P/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-Q8_K_P.gguf
1155+
sha256: b37a4225ba1eebaaac35b69dfeb78d1955d06fcb9f41fd334a4837cbdaff21c1
1156+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/Hermes3.6-35B-A3B-Uncensored-Genesis-V7-Q8_K_P.gguf
1157+
- filename: llama-cpp/mmproj/Hermes3.6-35B-A3B-Uncensored-Genesis-V7/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
1158+
sha256: c8e702344a81f8c226a914aa980ed6e1f604bce9374f1fed8e65c896908af414
1159+
uri: huggingface://LuffyTheFox/Qwen3.6-35B-A3B-Uncensored-Genesis-Hermes-V7-GGUF/mmproj-Hermes3.6-35B-A3B-Uncensored-Genesis-F16.gguf
9311160
- &kat-coder-v2-5-dev
9321161
name: "kat-coder-v2.5-dev"
9331162
variants:

0 commit comments

Comments
 (0)