gghfez commited on
Commit
4eb0d94
·
verified ·
1 Parent(s): 513622b

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +148 -2
README.md CHANGED
@@ -4,11 +4,157 @@ base_model:
4
  - neuronpedia/jacobian-lens
5
  ---
6
 
7
- For use with llama.cpp
8
 
9
- Tested / matches pytorch:
10
 
 
11
  Qwen-2.5-7b-Instruct
12
 
 
13
 
 
 
 
 
14
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  - neuronpedia/jacobian-lens
5
  ---
6
 
7
+ GGUF conversion of neuronpedia/jacobian-lens
8
 
9
+ Requires patches to work with llama.cpp
10
 
11
+ # Manually Tested / verified identical predictions
12
  Qwen-2.5-7b-Instruct
13
 
14
+ # Converted Models
15
 
16
+ ```
17
+ #gemma-2-27b
18
+ d_model=4608 n_prompts=780 layers=0..44 (45) dtype=f16
19
+ gguf: jacobian-lens-GGUF/gemma-2-27b-jlens.gguf
20
 
21
+ #gemma-2-2b
22
+ d_model=2304 n_prompts=454 layers=0..24 (25) dtype=f16
23
+ gguf: jacobian-lens-GGUF/gemma-2-2b-jlens.gguf
24
+
25
+ #gemma-2-2b-it
26
+ d_model=2304 n_prompts=337 layers=0..24 (25) dtype=f16
27
+ gguf: jacobian-lens-GGUF/gemma-2-2b-it-jlens.gguf
28
+
29
+ #gemma-2-9b
30
+ d_model=3584 n_prompts=462 layers=0..40 (41) dtype=f16
31
+ gguf: jacobian-lens-GGUF/gemma-2-9b-jlens.gguf
32
+
33
+ #gemma-2-9b-it
34
+ d_model=3584 n_prompts=453 layers=0..40 (41) dtype=f16
35
+ gguf: jacobian-lens-GGUF/gemma-2-9b-it-jlens.gguf
36
+
37
+ #gemma-3-12b
38
+ d_model=3840 n_prompts=775 layers=0..46 (47) dtype=f16
39
+ gguf: jacobian-lens-GGUF/gemma-3-12b-jlens.gguf
40
+
41
+ #gemma-3-12b-it
42
+ d_model=3840 n_prompts=844 layers=0..46 (47) dtype=f16
43
+ gguf: jacobian-lens-GGUF/gemma-3-12b-it-jlens.gguf
44
+
45
+ #gemma-3-1b
46
+ d_model=1152 n_prompts=467 layers=0..24 (25) dtype=f16
47
+ gguf: jacobian-lens-GGUF/gemma-3-1b-jlens.gguf
48
+
49
+ #gemma-3-1b-it
50
+ d_model=1152 n_prompts=460 layers=0..24 (25) dtype=f16
51
+ gguf: jacobian-lens-GGUF/gemma-3-1b-it-jlens.gguf
52
+
53
+ #gemma-3-270m
54
+ d_model=640 n_prompts=302 layers=0..16 (17) dtype=f16
55
+ gguf: jacobian-lens-GGUF/gemma-3-270m-jlens.gguf
56
+
57
+ #gemma-3-270m-it
58
+ d_model=640 n_prompts=278 layers=0..16 (17) dtype=f16
59
+ gguf: jacobian-lens-GGUF/gemma-3-270m-it-jlens.gguf
60
+
61
+ #gemma-3-27b
62
+ d_model=5376 n_prompts=670 layers=0..60 (61) dtype=f16
63
+ gguf: jacobian-lens-GGUF/gemma-3-27b-jlens.gguf
64
+
65
+ #gemma-3-27b-it
66
+ d_model=5376 n_prompts=828 layers=0..60 (61) dtype=f16
67
+ gguf: jacobian-lens-GGUF/gemma-3-27b-it-jlens.gguf
68
+
69
+ #gemma-3-4b
70
+ d_model=2560 n_prompts=616 layers=0..32 (33) dtype=f16
71
+ gguf: jacobian-lens-GGUF/gemma-3-4b-jlens.gguf
72
+
73
+ #gemma-3-4b-it
74
+ d_model=2560 n_prompts=546 layers=0..32 (33) dtype=f16
75
+ gguf: jacobian-lens-GGUF/gemma-3-4b-it-jlens.gguf
76
+
77
+ #gemma-4-31b
78
+ d_model=5376 n_prompts=861 layers=0..58 (59) dtype=f16
79
+ gguf: jacobian-lens-GGUF/gemma-4-31b-jlens.gguf
80
+
81
+ #gemma-4-e2b
82
+ d_model=1536 n_prompts=455 layers=0..33 (34) dtype=f16
83
+ gguf: jacobian-lens-GGUF/gemma-4-e2b-jlens.gguf
84
+
85
+ #gemma-4-e4b
86
+ d_model=2560 n_prompts=663 layers=0..40 (41) dtype=f16
87
+ gguf: jacobian-lens-GGUF/gemma-4-e4b-jlens.gguf
88
+
89
+ #gpt2-small
90
+ d_model=768 n_prompts=277 layers=0..10 (11) dtype=f16
91
+ gguf: jacobian-lens-GGUF/gpt2-small-jlens.gguf
92
+
93
+ #gpt-oss-20b
94
+ d_model=2880 n_prompts=457 layers=0..22 (23) dtype=f16
95
+ gguf: jacobian-lens-GGUF/gpt-oss-20b-jlens.gguf
96
+
97
+ #llama3.1-8b
98
+ d_model=4096 n_prompts=539 layers=0..30 (31) dtype=f16
99
+ gguf: jacobian-lens-GGUF/llama3.1-8b-jlens.gguf
100
+
101
+ #llama3.1-8b-it
102
+ d_model=4096 n_prompts=461 layers=0..30 (31) dtype=f16
103
+ gguf: jacobian-lens-GGUF/llama3.1-8b-it-jlens.gguf
104
+
105
+ #llama3.3-70b-it
106
+ d_model=8192 n_prompts=125 layers=0..78 (79) dtype=f16
107
+ gguf: jacobian-lens-GGUF/llama3.3-70b-it-jlens.gguf
108
+
109
+ #olmo-3-1025-7b
110
+ d_model=4096 n_prompts=616 layers=0..30 (31) dtype=f16
111
+ gguf: jacobian-lens-GGUF/olmo-3-1025-7b-jlens.gguf
112
+
113
+ #olmo-3-1125-32b
114
+ d_model=5120 n_prompts=470 layers=0..62 (63) dtype=f16
115
+ gguf: jacobian-lens-GGUF/olmo-3-1125-32b-jlens.gguf
116
+
117
+ #qwen2.5-7b-it
118
+ d_model=3584 n_prompts=485 layers=0..26 (27) dtype=f16
119
+ gguf: jacobian-lens-GGUF/qwen2.5-7b-it-jlens.gguf
120
+
121
+ #qwen3-14b
122
+ d_model=5120 n_prompts=615 layers=0..38 (39) dtype=f16
123
+ gguf: jacobian-lens-GGUF/qwen3-14b-jlens.gguf
124
+
125
+ #qwen3-1.7b
126
+ d_model=2048 n_prompts=466 layers=0..26 (27) dtype=f16
127
+ gguf: jacobian-lens-GGUF/qwen3-1.7b-jlens.gguf
128
+
129
+ #qwen3-4b
130
+ d_model=2560 n_prompts=479 layers=0..34 (35) dtype=f16
131
+ gguf: jacobian-lens-GGUF/qwen3-4b-jlens.gguf
132
+
133
+ #qwen3.5-0.8b
134
+ d_model=1024 n_prompts=233 layers=0..22 (23) dtype=f16
135
+ gguf: jacobian-lens-GGUF/qwen3.5-0.8b-jlens.gguf
136
+
137
+ #qwen3.5-27b
138
+ d_model=5120 n_prompts=672 layers=0..62 (63) dtype=f16
139
+ gguf: jacobian-lens-GGUF/qwen3.5-27b-jlens.gguf
140
+
141
+ #qwen3.5-2b-pt
142
+ d_model=2048 n_prompts=335 layers=0..22 (23) dtype=f16
143
+ gguf: jacobian-lens-GGUF/qwen3.5-2b-pt-jlens.gguf
144
+
145
+ #qwen3.5-4b
146
+ d_model=2560 n_prompts=1000 layers=0..30 (31) dtype=f16
147
+ gguf: jacobian-lens-GGUF/qwen3.5-4b-jlens.gguf
148
+
149
+ #qwen3.5-9b-pt
150
+ d_model=4096 n_prompts=458 layers=0..30 (31) dtype=f16
151
+ gguf: jacobian-lens-GGUF/qwen3.5-9b-pt-jlens.gguf
152
+
153
+ #qwen3.6-27b
154
+ d_model=5120 n_prompts=1000 layers=0..62 (63) dtype=f16
155
+ gguf: jacobian-lens-GGUF/qwen3.6-27b-jlens.gguf
156
+
157
+ #qwen3-8b
158
+ d_model=4096 n_prompts=461 layers=0..34 (35) dtype=f16
159
+ gguf: jacobian-lens-GGUF/qwen3-8b-jlens.gguf
160
+ ```