Views
No views yet
1- model: google/gemma-2-2b
2- model: google/gemma-2-2b-it
3 parameters:
4 density:
5 - filter: model.layers.1.self_attn.q_proj
6 value: 0.00539
7 - filter: model.layers.2.self_attn.q_proj
8 value: 0.03843
9 - filter: model.layers.6.self_attn.q_proj
10 value: 0.03716
11 - filter: model.layers.24.self_attn.q_proj
12 value: 0.04552
13 - filter: model.layers.25.self_attn.q_proj
14 value: 0.03919
15 - filter: model.layers.0.self_attn.k_proj
16 value: 0.00592
17 - filter: model.layers.2.self_attn.k_proj
18 value: 0.02603
19 - filter: model.layers.3.self_attn.k_proj
20 value: 0.07283
21 - filter: model.layers.8.self_attn.k_proj
22 value: 0.08753
23 - filter: model.layers.10.self_attn.k_proj
24 value: 0.07783
25 - filter: model.layers.23.self_attn.k_proj
26 value: 0.05987
27 - filter: model.layers.24.self_attn.k_proj
28 value: 0.02903
29 - filter: model.layers.25.self_attn.k_proj
30 value: 0.08715
31 - filter: model.layers.0.self_attn.v_proj
32 value: 0.03025
33 - filter: model.layers.2.self_attn.v_proj
34 value: 0.00286
35 - filter: model.layers.3.self_attn.v_proj
36 value: 0.09155
37 - filter: model.layers.6.self_attn.v_proj
38 value: 0.06811
39 - filter: model.layers.7.self_attn.v_proj
40 value: 0.01334
41 - filter: model.layers.10.self_attn.v_proj
42 value: 0.04
43 - filter: model.layers.13.self_attn.v_proj
44 value: 0.09347
45 - filter: model.layers.24.self_attn.v_proj
46 value: 0.07956
47 - filter: model.layers.0.self_attn.o_proj
48 value: 0.03265
49 - filter: model.layers.2.self_attn.o_proj
50 value: 0.06134
51 - filter: model.layers.4.self_attn.o_proj
52 value: 0.07924
53 - filter: model.layers.6.self_attn.o_proj
54 value: 0.09982
55 - filter: model.layers.7.self_attn.o_proj
56 value: 0.02826
57 - filter: model.layers.8.self_attn.o_proj
58 value: 0.03906
59 - filter: model.layers.19.self_attn.o_proj
60 value: 0.09507
61 - filter: model.layers.23.self_attn.o_proj
62 value: 0.00282
63 - filter: model.layers.24.self_attn.o_proj
64 value: 0.09864
65 - filter: model.layers.25.self_attn.o_proj
66 value: 0.00961
67 - filter: model.layers.1.mlp.gate_proj
68 value: 0.08775
69 - filter: model.layers.2.mlp.gate_proj
70 value: 0.0001
71 - filter: model.layers.6.mlp.gate_proj
72 value: 0.06577
73 - filter: model.layers.12.mlp.gate_proj
74 value: 0.02651
75 - filter: model.layers.13.mlp.gate_proj
76 value: 0.04687
77 - filter: model.layers.15.mlp.gate_proj
78 value: 0.03147
79 - filter: model.layers.16.mlp.gate_proj
80 value: 0.05726
81 - filter: model.layers.17.mlp.gate_proj
82 value: 0.04511
83 - filter: model.layers.23.mlp.gate_proj
84 value: 0.08641
85 - filter: model.layers.1.mlp.up_proj
86 value: 0.06887
87 - filter: model.layers.6.mlp.up_proj
88 value: 0.07411
89 - filter: model.layers.7.mlp.up_proj
90 value: 0.05424
91 - filter: model.layers.12.mlp.up_proj
92 value: 0.08044
93 - filter: model.layers.13.mlp.up_proj
94 value: 0.0021
95 - filter: model.layers.14.mlp.up_proj
96 value: 0.26389
97 - filter: model.layers.15.mlp.up_proj
98 value: 0.06886
99 - filter: model.layers.23.mlp.up_proj
100 value: 0.02931
101 - filter: model.layers.0.mlp.down_proj
102 value: 0.06756
103 - filter: model.layers.1.mlp.down_proj
104 value: 0.03746
105 - filter: model.layers.2.mlp.down_proj
106 value: 0.09104
107 - filter: model.layers.3.mlp.down_proj
108 value: 0.06643
109 - filter: model.layers.4.mlp.down_proj
110 value: 0.05003
111 - filter: model.layers.5.mlp.down_proj
112 value: 0.0406
113 - filter: model.layers.6.mlp.down_proj
114 value: 0.01609
115 - filter: model.layers.7.mlp.down_proj
116 value: 0.09629
117 - filter: model.layers.8.mlp.down_proj
118 value: 0.08912
119 - filter: model.layers.10.mlp.down_proj
120 value: 0.04635
121 - filter: model.layers.11.mlp.down_proj
122 value: 0.0099
123 - filter: model.layers.12.mlp.down_proj
124 value: 0.03487
125 - filter: model.layers.13.mlp.down_proj
126 value: 0.04977
127 - filter: model.layers.14.mlp.down_proj
128 value: 0.00393
129 - filter: model.layers.15.mlp.down_proj
130 value: 0.00748
131 - filter: model.layers.16.mlp.down_proj
132 value: 0.06696
133 - filter: model.layers.17.mlp.down_proj
134 value: 0.02067
135 - filter: model.layers.19.mlp.down_proj
136 value: 0.009
137 - filter: model.layers.20.mlp.down_proj
138 value: 0.0215
139 - filter: model.layers.21.mlp.down_proj
140 value: 0.04196
141 - filter: model.layers.22.mlp.down_proj
142 value: 0.06326
143 - filter: model.layers.25.mlp.down_proj
144 value: 0.04905
145 weight:
146 - value: 1
147merge_method: ties
148base_model: google/gemma-2-2b
149parameters:
150 normalize: true
151 int8_mask: true
152dtype: bfloat16
153tokenizer_source: union1!pip install -qU transformers accelerate
2
3from transformers import AutoTokenizer
4import transformers
5import torch
6
7model = "choprahetarth/gemma-instruct-merge-test_"
8messages = [{"role": "user", "content": "What is a large language model?"}]
9
10tokenizer = AutoTokenizer.from_pretrained(model)
11prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
12pipeline = transformers.pipeline(
13 "text-generation",
14 model=model,
15 torch_dtype=torch.float16,
16 device_map="auto",
17)
18
19outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
20print(outputs[0]["generated_text"])