Hugging Face's logo

OpenTransformer
/
unary-quantization-research

Safetensors
English
quantization
unary
thermometer-encoding
inference-engine
low-bit
Model card Files
xet
 Community
OpenTransformer commited on
Commit
50a9bcc
·
verified ·
1 Parent(s): 3137f27

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. deepseek-r1-1.5b-packed/config.json +13 -0
  2. deepseek-r1-1.5b-packed/model_layers_0_self_attn_q_proj_weight.scales +0 -0
  3. deepseek-r1-1.5b-packed/model_layers_10_mlp_up_proj_weight.scales +0 -0
  4. deepseek-r1-1.5b-packed/model_layers_10_self_attn_q_proj_bias.fp16 +0 -0
  5. deepseek-r1-1.5b-packed/model_layers_12_self_attn_k_proj_weight.scales +0 -0
  6. deepseek-r1-1.5b-packed/model_layers_12_self_attn_k_proj_weight.signs +0 -0
  7. deepseek-r1-1.5b-packed/model_layers_12_self_attn_o_proj_weight.scales +0 -0
  8. deepseek-r1-1.5b-packed/model_layers_14_self_attn_v_proj_weight.scales +0 -0
  9. deepseek-r1-1.5b-packed/model_layers_15_mlp_gate_proj_weight.scales +0 -0
  10. deepseek-r1-1.5b-packed/model_layers_15_self_attn_k_proj_weight.signs +0 -0
  11. deepseek-r1-1.5b-packed/model_layers_15_self_attn_q_proj_weight.rmm +1 -0
  12. deepseek-r1-1.5b-packed/model_layers_16_mlp_gate_proj_weight.scales +0 -0
  13. deepseek-r1-1.5b-packed/model_layers_16_self_attn_q_proj_weight.scales +0 -0
  14. deepseek-r1-1.5b-packed/model_layers_17_input_layernorm_weight.fp16 +0 -0
  15. deepseek-r1-1.5b-packed/model_layers_17_mlp_down_proj_weight.rmm +1 -0
  16. deepseek-r1-1.5b-packed/model_layers_17_mlp_up_proj_weight.rmm +1 -0
  17. deepseek-r1-1.5b-packed/model_layers_17_post_attention_layernorm_weight.fp16 +0 -0
  18. deepseek-r1-1.5b-packed/model_layers_17_self_attn_k_proj_weight.scales +0 -0
  19. deepseek-r1-1.5b-packed/model_layers_17_self_attn_q_proj_bias.fp16 +0 -0
  20. deepseek-r1-1.5b-packed/model_layers_18_self_attn_o_proj_weight.scales +0 -0
  21. deepseek-r1-1.5b-packed/model_layers_1_mlp_down_proj_weight.scales +0 -0
  22. deepseek-r1-1.5b-packed/model_layers_1_mlp_up_proj_weight.rmm +1 -0
  23. deepseek-r1-1.5b-packed/model_layers_1_self_attn_k_proj_weight.scales +0 -0
  24. deepseek-r1-1.5b-packed/model_layers_20_input_layernorm_weight.fp16 +0 -0
  25. deepseek-r1-1.5b-packed/model_layers_20_self_attn_q_proj_weight.rmm +1 -0
  26. deepseek-r1-1.5b-packed/model_layers_20_self_attn_v_proj_weight.rmm +1 -0
  27. deepseek-r1-1.5b-packed/model_layers_21_self_attn_q_proj_weight.rmm +1 -0
  28. deepseek-r1-1.5b-packed/model_layers_21_self_attn_q_proj_weight.scales +0 -0
  29. deepseek-r1-1.5b-packed/model_layers_22_mlp_down_proj_weight.rmm +1 -0
  30. deepseek-r1-1.5b-packed/model_layers_22_mlp_down_proj_weight.scales +0 -0
  31. deepseek-r1-1.5b-packed/model_layers_22_mlp_up_proj_weight.rmm +1 -0
  32. deepseek-r1-1.5b-packed/model_layers_22_self_attn_k_proj_weight.signs +0 -0
  33. deepseek-r1-1.5b-packed/model_layers_23_mlp_down_proj_weight.rmm +1 -0
  34. deepseek-r1-1.5b-packed/model_layers_23_mlp_gate_proj_weight.scales +0 -0
  35. deepseek-r1-1.5b-packed/model_layers_23_self_attn_o_proj_weight.scales +0 -0
  36. deepseek-r1-1.5b-packed/model_layers_24_mlp_gate_proj_weight.rmm +1 -0
  37. deepseek-r1-1.5b-packed/model_layers_25_self_attn_q_proj_bias.fp16 +0 -0
  38. deepseek-r1-1.5b-packed/model_layers_26_self_attn_v_proj_bias.fp16 +0 -0
  39. deepseek-r1-1.5b-packed/model_layers_27_self_attn_v_proj_weight.scales +0 -0
  40. deepseek-r1-1.5b-packed/model_layers_27_self_attn_v_proj_weight.signs +0 -0
  41. deepseek-r1-1.5b-packed/model_layers_3_self_attn_q_proj_weight.rmm +1 -0
  42. deepseek-r1-1.5b-packed/model_layers_3_self_attn_q_proj_weight.scales +0 -0
  43. deepseek-r1-1.5b-packed/model_layers_3_self_attn_v_proj_weight.scales +0 -0
  44. deepseek-r1-1.5b-packed/model_layers_4_self_attn_v_proj_weight.signs +0 -0
  45. deepseek-r1-1.5b-packed/model_layers_5_self_attn_k_proj_bias.fp16 +0 -0
  46. deepseek-r1-1.5b-packed/model_layers_5_self_attn_k_proj_weight.scales +0 -0
  47. deepseek-r1-1.5b-packed/model_layers_5_self_attn_o_proj_weight.rmm +1 -0
  48. deepseek-r1-1.5b-packed/model_layers_5_self_attn_o_proj_weight.scales +0 -0
  49. deepseek-r1-1.5b-packed/model_layers_6_mlp_down_proj_weight.rmm +1 -0
  50. deepseek-r1-1.5b-packed/model_layers_7_mlp_up_proj_weight.rmm +1 -0
deepseek-r1-1.5b-packed/config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "n_levels": 7,
12
+ "quant_type": "packed_unary"
13
+ }
deepseek-r1-1.5b-packed/model_layers_0_self_attn_q_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_10_mlp_up_proj_weight.scales ADDED
Binary file (35.8 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_10_self_attn_q_proj_bias.fp16 ADDED
Binary file (3.07 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_12_self_attn_k_proj_weight.scales ADDED
Binary file (1.02 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_12_self_attn_k_proj_weight.signs ADDED
Binary file (49.2 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_12_self_attn_o_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_14_self_attn_v_proj_weight.scales ADDED
Binary file (1.02 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_15_mlp_gate_proj_weight.scales ADDED
Binary file (35.8 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_15_self_attn_k_proj_weight.signs ADDED
Binary file (49.2 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_15_self_attn_q_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_16_mlp_gate_proj_weight.scales ADDED
Binary file (35.8 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_16_self_attn_q_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_17_input_layernorm_weight.fp16 ADDED
Binary file (3.07 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_17_mlp_down_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_17_mlp_up_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_17_post_attention_layernorm_weight.fp16 ADDED
Binary file (3.07 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_17_self_attn_k_proj_weight.scales ADDED
Binary file (1.02 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_17_self_attn_q_proj_bias.fp16 ADDED
Binary file (3.07 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_18_self_attn_o_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_1_mlp_down_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_1_mlp_up_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_1_self_attn_k_proj_weight.scales ADDED
Binary file (1.02 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_20_input_layernorm_weight.fp16 ADDED
Binary file (3.07 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_20_self_attn_q_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_20_self_attn_v_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_21_self_attn_q_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_21_self_attn_q_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_22_mlp_down_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_22_mlp_down_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_22_mlp_up_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_22_self_attn_k_proj_weight.signs ADDED
Binary file (49.2 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_23_mlp_down_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_23_mlp_gate_proj_weight.scales ADDED
Binary file (35.8 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_23_self_attn_o_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_24_mlp_gate_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_25_self_attn_q_proj_bias.fp16 ADDED
Binary file (3.07 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_26_self_attn_v_proj_bias.fp16 ADDED
Binary file (512 Bytes). View file
 
deepseek-r1-1.5b-packed/model_layers_27_self_attn_v_proj_weight.scales ADDED
Binary file (1.02 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_27_self_attn_v_proj_weight.signs ADDED
Binary file (49.2 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_3_self_attn_q_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_3_self_attn_q_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_3_self_attn_v_proj_weight.scales ADDED
Binary file (1.02 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_4_self_attn_v_proj_weight.signs ADDED
Binary file (49.2 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_5_self_attn_k_proj_bias.fp16 ADDED
Binary file (512 Bytes). View file
 
deepseek-r1-1.5b-packed/model_layers_5_self_attn_k_proj_weight.scales ADDED
Binary file (1.02 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_5_self_attn_o_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_5_self_attn_o_proj_weight.scales ADDED
Binary file (6.14 kB). View file
 
deepseek-r1-1.5b-packed/model_layers_6_mlp_down_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+ 
deepseek-r1-1.5b-packed/model_layers_7_mlp_up_proj_weight.rmm ADDED
@@ -0,0 +1 @@
 
 
1
+