Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- deepseek-r1-1.5b-ternary/config.json +12 -0
- deepseek-r1-1.5b-ternary/model_layers_0_self_attn_q_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_0_self_attn_v_proj_weight.neg +0 -0
- deepseek-r1-1.5b-ternary/model_layers_10_self_attn_k_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_11_self_attn_k_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_12_self_attn_k_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_12_self_attn_o_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_13_input_layernorm_weight.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_13_self_attn_k_proj_weight.scales +8 -0
- deepseek-r1-1.5b-ternary/model_layers_13_self_attn_v_proj_weight.pos +0 -0
- deepseek-r1-1.5b-ternary/model_layers_14_self_attn_q_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_15_mlp_gate_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_16_mlp_down_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_16_mlp_gate_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_16_self_attn_q_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_16_self_attn_v_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_16_self_attn_v_proj_weight.neg +0 -0
- deepseek-r1-1.5b-ternary/model_layers_17_input_layernorm_weight.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_17_post_attention_layernorm_weight.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_17_self_attn_k_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_17_self_attn_q_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_17_self_attn_v_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_18_input_layernorm_weight.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_18_mlp_up_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_18_self_attn_o_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_19_self_attn_v_proj_weight.pos +0 -0
- deepseek-r1-1.5b-ternary/model_layers_1_mlp_down_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_1_self_attn_k_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_1_self_attn_q_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_20_input_layernorm_weight.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_20_self_attn_q_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_20_self_attn_v_proj_weight.pos +0 -0
- deepseek-r1-1.5b-ternary/model_layers_21_mlp_up_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_21_post_attention_layernorm_weight.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_21_self_attn_k_proj_weight.neg +0 -0
- deepseek-r1-1.5b-ternary/model_layers_21_self_attn_q_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_22_mlp_down_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_22_mlp_gate_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_23_input_layernorm_weight.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_23_mlp_gate_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_23_self_attn_o_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_23_self_attn_v_proj_weight.neg +0 -0
- deepseek-r1-1.5b-ternary/model_layers_24_self_attn_v_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_24_self_attn_v_proj_weight.neg +0 -0
- deepseek-r1-1.5b-ternary/model_layers_25_self_attn_q_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_26_mlp_gate_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_26_self_attn_k_proj_weight.pos +0 -0
- deepseek-r1-1.5b-ternary/model_layers_26_self_attn_v_proj_bias.fp16 +0 -0
- deepseek-r1-1.5b-ternary/model_layers_27_self_attn_q_proj_weight.scales +0 -0
- deepseek-r1-1.5b-ternary/model_layers_27_self_attn_v_proj_weight.pos +0 -0
deepseek-r1-1.5b-ternary/config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"hidden_size": 1536,
|
| 3 |
+
"intermediate_size": 8960,
|
| 4 |
+
"num_attention_heads": 12,
|
| 5 |
+
"num_key_value_heads": 2,
|
| 6 |
+
"num_hidden_layers": 28,
|
| 7 |
+
"vocab_size": 151936,
|
| 8 |
+
"head_dim": 128,
|
| 9 |
+
"rope_theta": 1000000.0,
|
| 10 |
+
"rms_norm_eps": 1e-06,
|
| 11 |
+
"alpha": 0.7
|
| 12 |
+
}
|
deepseek-r1-1.5b-ternary/model_layers_0_self_attn_q_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_0_self_attn_v_proj_weight.neg
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_10_self_attn_k_proj_weight.scales
ADDED
|
Binary file (1.02 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_11_self_attn_k_proj_bias.fp16
ADDED
|
Binary file (512 Bytes). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_12_self_attn_k_proj_weight.scales
ADDED
|
Binary file (1.02 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_12_self_attn_o_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_13_input_layernorm_weight.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_13_self_attn_k_proj_weight.scales
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
s��<���<��<�k6=�c�<[�;=w�<M[8=���<�Ф<&R7=�-1=}�=yI�<�n=�a(=U--=��=�;=o�F=ML�<c}S=�d=^T=��#=�nO=j�
|
| 2 |
+
=�P=Ӆ?=;.A=�Z=�r:=��b=CQg=�Z=��d=�}=Ofe=1k=Ah=%�c=�^p="�j=E�c={_=��p=��v=/�m=��r=�wp=��t=�n=.A�<~jg=a=��T=��X=(=Z=�!V=��R=�<�#`=1�V=49O=�
|
| 3 |
+
Y<���<@��<M�!=b��<
|
| 4 |
+
��<g�=]�'=U�B=���<
|
| 5 |
+
=�3_=L6X=`=��W="?=��i=N�i=�i=_h=M�^=?�p=��v=(�]=�n=7�k=Z7y=p�q=y�p=��`=]�s=�P=�
|
| 6 |
+
]=(wd=^�k=��]=5Vb=�ji=߅Q=��l=4L^=�H=[
|
| 7 |
+
M=��=���<�=�a�<�H'=A�<\
|
| 8 |
+
1=&�=���<i?*=F�9=���<)�D=s�=�)R=��C=e>W=9Y=ѠN=}�g=Y,I=��f=�'q=��J=\�=�!h=H�=�
|
deepseek-r1-1.5b-ternary/model_layers_13_self_attn_v_proj_weight.pos
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_14_self_attn_q_proj_bias.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_15_mlp_gate_proj_weight.scales
ADDED
|
Binary file (35.8 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_16_mlp_down_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_16_mlp_gate_proj_weight.scales
ADDED
|
Binary file (35.8 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_16_self_attn_q_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_16_self_attn_v_proj_bias.fp16
ADDED
|
Binary file (512 Bytes). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_16_self_attn_v_proj_weight.neg
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_17_input_layernorm_weight.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_17_post_attention_layernorm_weight.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_17_self_attn_k_proj_weight.scales
ADDED
|
Binary file (1.02 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_17_self_attn_q_proj_bias.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_17_self_attn_v_proj_weight.scales
ADDED
|
Binary file (1.02 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_18_input_layernorm_weight.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_18_mlp_up_proj_weight.scales
ADDED
|
Binary file (35.8 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_18_self_attn_o_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_19_self_attn_v_proj_weight.pos
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_1_mlp_down_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_1_self_attn_k_proj_weight.scales
ADDED
|
Binary file (1.02 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_1_self_attn_q_proj_bias.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_20_input_layernorm_weight.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_20_self_attn_q_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_20_self_attn_v_proj_weight.pos
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_21_mlp_up_proj_weight.scales
ADDED
|
Binary file (35.8 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_21_post_attention_layernorm_weight.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_21_self_attn_k_proj_weight.neg
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_21_self_attn_q_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_22_mlp_down_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_22_mlp_gate_proj_weight.scales
ADDED
|
Binary file (35.8 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_23_input_layernorm_weight.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_23_mlp_gate_proj_weight.scales
ADDED
|
Binary file (35.8 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_23_self_attn_o_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_23_self_attn_v_proj_weight.neg
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_24_self_attn_v_proj_bias.fp16
ADDED
|
Binary file (512 Bytes). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_24_self_attn_v_proj_weight.neg
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_25_self_attn_q_proj_bias.fp16
ADDED
|
Binary file (3.07 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_26_mlp_gate_proj_weight.scales
ADDED
|
Binary file (35.8 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_26_self_attn_k_proj_weight.pos
ADDED
|
Binary file (49.2 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_26_self_attn_v_proj_bias.fp16
ADDED
|
Binary file (512 Bytes). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_27_self_attn_q_proj_weight.scales
ADDED
|
Binary file (6.14 kB). View file
|
|
|
deepseek-r1-1.5b-ternary/model_layers_27_self_attn_v_proj_weight.pos
ADDED
|
Binary file (49.2 kB). View file
|
|
|