@bendyline/gilde 0.1.56 → 0.1.57
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/authoring/chat-models/qwen3.8-flash-next-iq3.json +198 -0
- package/authoring/chat-models/qwen3.8-flash-next-iq4.json +198 -0
- package/data/chat-models/index.json +1 -1
- package/data/chat-models/qw/qwen3.8-flash-next-iq3/manifest.json +221 -0
- package/data/chat-models/qw/qwen3.8-flash-next-iq3/versions/1.0.0/manifest.json +36 -0
- package/data/chat-models/qw/qwen3.8-flash-next-iq4/manifest.json +221 -0
- package/data/chat-models/qw/qwen3.8-flash-next-iq4/versions/1.0.0/manifest.json +36 -0
- package/package.json +1 -1
|
@@ -0,0 +1,198 @@
|
|
|
1
|
+
{
|
|
2
|
+
"id": "qwen3.8-flash-next-iq3",
|
|
3
|
+
"name": "Qwen 3.8 Flash Next (180B-A6B, IQ3_XXS)",
|
|
4
|
+
"description": "Qwen's experimental preview of the architecture behind Qwen4: a 125B-parameter MoE with 6B active per token, plus a 51B n-gram embedding table designed for cheap offload. This Unsloth UD-IQ3_XXS build is an 82 GB split GGUF whose 26.8 GiB n-gram table can be demand-paged; it is the practical variant for 64 GB system RAM plus a 32 GB-class discrete GPU. Gezel starts it conservatively at 64K context and one slot, then chooses the expert GPU/CPU split from the machine's measured VRAM.",
|
|
5
|
+
"tags": [
|
|
6
|
+
"qwen",
|
|
7
|
+
"agentic",
|
|
8
|
+
"coding",
|
|
9
|
+
"multimodal",
|
|
10
|
+
"vision",
|
|
11
|
+
"video",
|
|
12
|
+
"tools",
|
|
13
|
+
"reasoning",
|
|
14
|
+
"long-context",
|
|
15
|
+
"moe",
|
|
16
|
+
"ngram-embedding",
|
|
17
|
+
"large",
|
|
18
|
+
"experimental"
|
|
19
|
+
],
|
|
20
|
+
"category": "general",
|
|
21
|
+
"maintainer": {
|
|
22
|
+
"name": "Unsloth",
|
|
23
|
+
"url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF"
|
|
24
|
+
},
|
|
25
|
+
"maker": {
|
|
26
|
+
"name": "Qwen",
|
|
27
|
+
"url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next"
|
|
28
|
+
},
|
|
29
|
+
"version": "1.0.0",
|
|
30
|
+
"versionMinGezelVersion": "1.26248",
|
|
31
|
+
"updatedAt": "2026-09-06T00:00:00Z",
|
|
32
|
+
"license": "qwen-community-1.0",
|
|
33
|
+
"licenseClass": "custom-restricted",
|
|
34
|
+
"licenseShortName": "Qwen Community 1.0",
|
|
35
|
+
"licenseUrl": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/blob/main/LICENSE",
|
|
36
|
+
"parameterSize": "180B",
|
|
37
|
+
"approxSizeBytes": 81961823936,
|
|
38
|
+
"supportsTools": true,
|
|
39
|
+
"contextWindow": 262144,
|
|
40
|
+
"upstream": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
|
|
41
|
+
"style": {
|
|
42
|
+
"family": "qwen",
|
|
43
|
+
"reasoningFormat": "think",
|
|
44
|
+
"toolCallFormat": "function-call"
|
|
45
|
+
},
|
|
46
|
+
"behaviors": [
|
|
47
|
+
"reasoning.strip-think-tags",
|
|
48
|
+
"provider.merge-system-messages",
|
|
49
|
+
"mcp.compact-tool-schemas",
|
|
50
|
+
"provider.compact-write-transcript",
|
|
51
|
+
"fabrication.detect-past-tense-no-tools",
|
|
52
|
+
"turn.ollama-num-predict-bumped",
|
|
53
|
+
"turn.preamble-folding",
|
|
54
|
+
"turn.ramble-detection",
|
|
55
|
+
"tools.mlx-grammar",
|
|
56
|
+
"prompt.retrieval-first",
|
|
57
|
+
"prompt.workspace-gestalt",
|
|
58
|
+
"prompt.derive-by-execution"
|
|
59
|
+
],
|
|
60
|
+
"tuning": {
|
|
61
|
+
"sampling": {
|
|
62
|
+
"temperature": 1,
|
|
63
|
+
"topP": 0.95,
|
|
64
|
+
"topK": 20,
|
|
65
|
+
"minP": 0,
|
|
66
|
+
"presencePenalty": 0,
|
|
67
|
+
"repetitionPenalty": 1,
|
|
68
|
+
"maxTokens": 12288
|
|
69
|
+
},
|
|
70
|
+
"samplingWhenThinking": {
|
|
71
|
+
"temperature": 1,
|
|
72
|
+
"topP": 0.95
|
|
73
|
+
},
|
|
74
|
+
"reasoning": {
|
|
75
|
+
"enableThinking": true,
|
|
76
|
+
"thinkingBudget": 4096,
|
|
77
|
+
"templateKwargs": {
|
|
78
|
+
"reasoning_effort": "xhigh"
|
|
79
|
+
}
|
|
80
|
+
},
|
|
81
|
+
"promptTags": {
|
|
82
|
+
"enableThinkingTag": "/think",
|
|
83
|
+
"disableThinkingTag": "/no_think"
|
|
84
|
+
},
|
|
85
|
+
"profiles": {
|
|
86
|
+
"thinking-general": {
|
|
87
|
+
"sampling": {
|
|
88
|
+
"temperature": 1,
|
|
89
|
+
"topP": 0.95,
|
|
90
|
+
"topK": 20,
|
|
91
|
+
"minP": 0,
|
|
92
|
+
"presencePenalty": 0,
|
|
93
|
+
"repetitionPenalty": 1,
|
|
94
|
+
"maxTokens": 12288
|
|
95
|
+
},
|
|
96
|
+
"reasoning": {
|
|
97
|
+
"enableThinking": true,
|
|
98
|
+
"thinkingBudget": 4096,
|
|
99
|
+
"templateKwargs": {
|
|
100
|
+
"reasoning_effort": "xhigh"
|
|
101
|
+
}
|
|
102
|
+
}
|
|
103
|
+
},
|
|
104
|
+
"thinking-deep": {
|
|
105
|
+
"reasoning": {
|
|
106
|
+
"enableThinking": true,
|
|
107
|
+
"thinkingBudget": 4096,
|
|
108
|
+
"templateKwargs": {
|
|
109
|
+
"reasoning_effort": "xhigh"
|
|
110
|
+
}
|
|
111
|
+
}
|
|
112
|
+
},
|
|
113
|
+
"thinking-coding": {
|
|
114
|
+
"sampling": {
|
|
115
|
+
"temperature": 1,
|
|
116
|
+
"topP": 0.95,
|
|
117
|
+
"topK": 20,
|
|
118
|
+
"minP": 0,
|
|
119
|
+
"presencePenalty": 0,
|
|
120
|
+
"repetitionPenalty": 1,
|
|
121
|
+
"maxTokens": 16384
|
|
122
|
+
},
|
|
123
|
+
"reasoning": {
|
|
124
|
+
"enableThinking": true,
|
|
125
|
+
"thinkingBudget": 4096,
|
|
126
|
+
"templateKwargs": {
|
|
127
|
+
"reasoning_effort": "xhigh"
|
|
128
|
+
}
|
|
129
|
+
}
|
|
130
|
+
},
|
|
131
|
+
"thinking-precise": {
|
|
132
|
+
"sampling": {
|
|
133
|
+
"temperature": 0.6,
|
|
134
|
+
"topP": 0.95,
|
|
135
|
+
"topK": 20,
|
|
136
|
+
"minP": 0,
|
|
137
|
+
"presencePenalty": 0,
|
|
138
|
+
"repetitionPenalty": 1,
|
|
139
|
+
"seed": 0,
|
|
140
|
+
"maxTokens": 6144
|
|
141
|
+
},
|
|
142
|
+
"reasoning": {
|
|
143
|
+
"enableThinking": true,
|
|
144
|
+
"thinkingBudget": 4096,
|
|
145
|
+
"templateKwargs": {
|
|
146
|
+
"reasoning_effort": "medium"
|
|
147
|
+
}
|
|
148
|
+
}
|
|
149
|
+
},
|
|
150
|
+
"instruct": {
|
|
151
|
+
"sampling": {
|
|
152
|
+
"temperature": 0.7,
|
|
153
|
+
"topP": 0.8,
|
|
154
|
+
"topK": 20,
|
|
155
|
+
"minP": 0,
|
|
156
|
+
"presencePenalty": 1.5,
|
|
157
|
+
"repetitionPenalty": 1,
|
|
158
|
+
"maxTokens": 4096
|
|
159
|
+
},
|
|
160
|
+
"reasoning": {
|
|
161
|
+
"enableThinking": false
|
|
162
|
+
}
|
|
163
|
+
},
|
|
164
|
+
"creative": {
|
|
165
|
+
"sampling": {
|
|
166
|
+
"temperature": 1,
|
|
167
|
+
"topP": 0.95,
|
|
168
|
+
"topK": 40,
|
|
169
|
+
"minP": 0,
|
|
170
|
+
"presencePenalty": 1.5,
|
|
171
|
+
"repetitionPenalty": 1,
|
|
172
|
+
"maxTokens": 8192
|
|
173
|
+
},
|
|
174
|
+
"reasoning": {
|
|
175
|
+
"enableThinking": false
|
|
176
|
+
}
|
|
177
|
+
}
|
|
178
|
+
},
|
|
179
|
+
"engine": {
|
|
180
|
+
"llamaCpp": {
|
|
181
|
+
"loadMode": "auto",
|
|
182
|
+
"lazyMode": "auto",
|
|
183
|
+
"cacheReuse": 0,
|
|
184
|
+
"flashAttn": "on",
|
|
185
|
+
"batchSize": 512,
|
|
186
|
+
"ubatchSize": 512,
|
|
187
|
+
"contextSize": 65536
|
|
188
|
+
}
|
|
189
|
+
}
|
|
190
|
+
},
|
|
191
|
+
"llamaCpp": {
|
|
192
|
+
"huggingfaceRepo": "unsloth/Qwen3.8-Flash-Next-GGUF",
|
|
193
|
+
"shardsDir": "UD-IQ3_XXS",
|
|
194
|
+
"quantization": "UD-IQ3_XXS",
|
|
195
|
+
"residentBytes": 53150661120,
|
|
196
|
+
"mmprojFilename": "mmproj-F16.gguf"
|
|
197
|
+
}
|
|
198
|
+
}
|
|
@@ -0,0 +1,198 @@
|
|
|
1
|
+
{
|
|
2
|
+
"id": "qwen3.8-flash-next-iq4",
|
|
3
|
+
"name": "Qwen 3.8 Flash Next (180B-A6B, IQ4_XS)",
|
|
4
|
+
"description": "Qwen's experimental preview of the architecture behind Qwen4: a 125B-parameter MoE with 6B active per token, plus a 51B n-gram embedding table designed for cheap offload. This higher-fidelity Unsloth UD-IQ4_XS build is a 94 GB split GGUF whose 26.8 GiB n-gram table can be demand-paged. It needs roughly 60.4 GiB for resident weights before KV cache and runtime overhead, so 96 GB system RAM is recommended; a 64 GB system with a 32 GB discrete GPU is possible but leaves almost no operating headroom.",
|
|
5
|
+
"tags": [
|
|
6
|
+
"qwen",
|
|
7
|
+
"agentic",
|
|
8
|
+
"coding",
|
|
9
|
+
"multimodal",
|
|
10
|
+
"vision",
|
|
11
|
+
"video",
|
|
12
|
+
"tools",
|
|
13
|
+
"reasoning",
|
|
14
|
+
"long-context",
|
|
15
|
+
"moe",
|
|
16
|
+
"ngram-embedding",
|
|
17
|
+
"large",
|
|
18
|
+
"experimental"
|
|
19
|
+
],
|
|
20
|
+
"category": "general",
|
|
21
|
+
"maintainer": {
|
|
22
|
+
"name": "Unsloth",
|
|
23
|
+
"url": "https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF"
|
|
24
|
+
},
|
|
25
|
+
"maker": {
|
|
26
|
+
"name": "Qwen",
|
|
27
|
+
"url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next"
|
|
28
|
+
},
|
|
29
|
+
"version": "1.0.0",
|
|
30
|
+
"versionMinGezelVersion": "1.26248",
|
|
31
|
+
"updatedAt": "2026-09-06T00:00:00Z",
|
|
32
|
+
"license": "qwen-community-1.0",
|
|
33
|
+
"licenseClass": "custom-restricted",
|
|
34
|
+
"licenseShortName": "Qwen Community 1.0",
|
|
35
|
+
"licenseUrl": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next/blob/main/LICENSE",
|
|
36
|
+
"parameterSize": "180B",
|
|
37
|
+
"approxSizeBytes": 93682584224,
|
|
38
|
+
"supportsTools": true,
|
|
39
|
+
"contextWindow": 262144,
|
|
40
|
+
"upstream": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
|
|
41
|
+
"style": {
|
|
42
|
+
"family": "qwen",
|
|
43
|
+
"reasoningFormat": "think",
|
|
44
|
+
"toolCallFormat": "function-call"
|
|
45
|
+
},
|
|
46
|
+
"behaviors": [
|
|
47
|
+
"reasoning.strip-think-tags",
|
|
48
|
+
"provider.merge-system-messages",
|
|
49
|
+
"mcp.compact-tool-schemas",
|
|
50
|
+
"provider.compact-write-transcript",
|
|
51
|
+
"fabrication.detect-past-tense-no-tools",
|
|
52
|
+
"turn.ollama-num-predict-bumped",
|
|
53
|
+
"turn.preamble-folding",
|
|
54
|
+
"turn.ramble-detection",
|
|
55
|
+
"tools.mlx-grammar",
|
|
56
|
+
"prompt.retrieval-first",
|
|
57
|
+
"prompt.workspace-gestalt",
|
|
58
|
+
"prompt.derive-by-execution"
|
|
59
|
+
],
|
|
60
|
+
"tuning": {
|
|
61
|
+
"sampling": {
|
|
62
|
+
"temperature": 1,
|
|
63
|
+
"topP": 0.95,
|
|
64
|
+
"topK": 20,
|
|
65
|
+
"minP": 0,
|
|
66
|
+
"presencePenalty": 0,
|
|
67
|
+
"repetitionPenalty": 1,
|
|
68
|
+
"maxTokens": 12288
|
|
69
|
+
},
|
|
70
|
+
"samplingWhenThinking": {
|
|
71
|
+
"temperature": 1,
|
|
72
|
+
"topP": 0.95
|
|
73
|
+
},
|
|
74
|
+
"reasoning": {
|
|
75
|
+
"enableThinking": true,
|
|
76
|
+
"thinkingBudget": 4096,
|
|
77
|
+
"templateKwargs": {
|
|
78
|
+
"reasoning_effort": "xhigh"
|
|
79
|
+
}
|
|
80
|
+
},
|
|
81
|
+
"promptTags": {
|
|
82
|
+
"enableThinkingTag": "/think",
|
|
83
|
+
"disableThinkingTag": "/no_think"
|
|
84
|
+
},
|
|
85
|
+
"profiles": {
|
|
86
|
+
"thinking-general": {
|
|
87
|
+
"sampling": {
|
|
88
|
+
"temperature": 1,
|
|
89
|
+
"topP": 0.95,
|
|
90
|
+
"topK": 20,
|
|
91
|
+
"minP": 0,
|
|
92
|
+
"presencePenalty": 0,
|
|
93
|
+
"repetitionPenalty": 1,
|
|
94
|
+
"maxTokens": 12288
|
|
95
|
+
},
|
|
96
|
+
"reasoning": {
|
|
97
|
+
"enableThinking": true,
|
|
98
|
+
"thinkingBudget": 4096,
|
|
99
|
+
"templateKwargs": {
|
|
100
|
+
"reasoning_effort": "xhigh"
|
|
101
|
+
}
|
|
102
|
+
}
|
|
103
|
+
},
|
|
104
|
+
"thinking-deep": {
|
|
105
|
+
"reasoning": {
|
|
106
|
+
"enableThinking": true,
|
|
107
|
+
"thinkingBudget": 4096,
|
|
108
|
+
"templateKwargs": {
|
|
109
|
+
"reasoning_effort": "xhigh"
|
|
110
|
+
}
|
|
111
|
+
}
|
|
112
|
+
},
|
|
113
|
+
"thinking-coding": {
|
|
114
|
+
"sampling": {
|
|
115
|
+
"temperature": 1,
|
|
116
|
+
"topP": 0.95,
|
|
117
|
+
"topK": 20,
|
|
118
|
+
"minP": 0,
|
|
119
|
+
"presencePenalty": 0,
|
|
120
|
+
"repetitionPenalty": 1,
|
|
121
|
+
"maxTokens": 16384
|
|
122
|
+
},
|
|
123
|
+
"reasoning": {
|
|
124
|
+
"enableThinking": true,
|
|
125
|
+
"thinkingBudget": 4096,
|
|
126
|
+
"templateKwargs": {
|
|
127
|
+
"reasoning_effort": "xhigh"
|
|
128
|
+
}
|
|
129
|
+
}
|
|
130
|
+
},
|
|
131
|
+
"thinking-precise": {
|
|
132
|
+
"sampling": {
|
|
133
|
+
"temperature": 0.6,
|
|
134
|
+
"topP": 0.95,
|
|
135
|
+
"topK": 20,
|
|
136
|
+
"minP": 0,
|
|
137
|
+
"presencePenalty": 0,
|
|
138
|
+
"repetitionPenalty": 1,
|
|
139
|
+
"seed": 0,
|
|
140
|
+
"maxTokens": 6144
|
|
141
|
+
},
|
|
142
|
+
"reasoning": {
|
|
143
|
+
"enableThinking": true,
|
|
144
|
+
"thinkingBudget": 4096,
|
|
145
|
+
"templateKwargs": {
|
|
146
|
+
"reasoning_effort": "medium"
|
|
147
|
+
}
|
|
148
|
+
}
|
|
149
|
+
},
|
|
150
|
+
"instruct": {
|
|
151
|
+
"sampling": {
|
|
152
|
+
"temperature": 0.7,
|
|
153
|
+
"topP": 0.8,
|
|
154
|
+
"topK": 20,
|
|
155
|
+
"minP": 0,
|
|
156
|
+
"presencePenalty": 1.5,
|
|
157
|
+
"repetitionPenalty": 1,
|
|
158
|
+
"maxTokens": 4096
|
|
159
|
+
},
|
|
160
|
+
"reasoning": {
|
|
161
|
+
"enableThinking": false
|
|
162
|
+
}
|
|
163
|
+
},
|
|
164
|
+
"creative": {
|
|
165
|
+
"sampling": {
|
|
166
|
+
"temperature": 1,
|
|
167
|
+
"topP": 0.95,
|
|
168
|
+
"topK": 40,
|
|
169
|
+
"minP": 0,
|
|
170
|
+
"presencePenalty": 1.5,
|
|
171
|
+
"repetitionPenalty": 1,
|
|
172
|
+
"maxTokens": 8192
|
|
173
|
+
},
|
|
174
|
+
"reasoning": {
|
|
175
|
+
"enableThinking": false
|
|
176
|
+
}
|
|
177
|
+
}
|
|
178
|
+
},
|
|
179
|
+
"engine": {
|
|
180
|
+
"llamaCpp": {
|
|
181
|
+
"loadMode": "auto",
|
|
182
|
+
"lazyMode": "auto",
|
|
183
|
+
"cacheReuse": 0,
|
|
184
|
+
"flashAttn": "on",
|
|
185
|
+
"batchSize": 512,
|
|
186
|
+
"ubatchSize": 512,
|
|
187
|
+
"contextSize": 65536
|
|
188
|
+
}
|
|
189
|
+
}
|
|
190
|
+
},
|
|
191
|
+
"llamaCpp": {
|
|
192
|
+
"huggingfaceRepo": "unsloth/Qwen3.8-Flash-Next-GGUF",
|
|
193
|
+
"shardsDir": "UD-IQ4_XS",
|
|
194
|
+
"quantization": "UD-IQ4_XS",
|
|
195
|
+
"residentBytes": 64871421440,
|
|
196
|
+
"mmprojFilename": "mmproj-F16.gguf"
|
|
197
|
+
}
|
|
198
|
+
}
|