Hugging Face's logo

bigscience
/
mt0-t5x

Model card Files
xet
Community
TimeRobber commited on
Commit
81d9a5a
·
1 Parent(s): a6bfc26

Upload part 0

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +45 -0
  2. checkpoint_1007000/checkpoint +3 -0
  3. checkpoint_1007000/state.param_states.decoder.layers_14.pre_mlp_layer_norm.scale.v/.zarray +1 -0
  4. checkpoint_1007000/state.param_states.decoder.layers_14.pre_mlp_layer_norm.scale.v/0 +0 -0
  5. checkpoint_1007000/state.param_states.decoder.layers_19.pre_self_attention_layer_norm.scale.v/.zarray +1 -0
  6. checkpoint_1007000/state.param_states.decoder.layers_19.pre_self_attention_layer_norm.scale.v/0 +0 -0
  7. checkpoint_1007000/state.param_states.decoder.layers_5.pre_mlp_layer_norm.scale.v/.zarray +1 -0
  8. checkpoint_1007000/state.param_states.decoder.layers_5.pre_mlp_layer_norm.scale.v/0 +0 -0
  9. checkpoint_1007000/state.param_states.decoder.layers_7.pre_mlp_layer_norm.scale.v/.zarray +1 -0
  10. checkpoint_1007000/state.param_states.decoder.layers_7.pre_mlp_layer_norm.scale.v/0 +0 -0
  11. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray +1 -0
  12. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp11477047902703534671.~1663960300386635~ +1 -0
  13. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp2065349081025687113.~1663960300549585~ +1 -0
  14. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp444933188802333506.~1663960300873156~ +1 -0
  15. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp4798233591127119695.~1663960300685739~ +1 -0
  16. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp5336578133756373591.~1663960300505112~ +1 -0
  17. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp8112336193798835172.~1663960300777247~ +1 -0
  18. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp9467005472495202151.~1663960300500527~ +1 -0
  19. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/0.0 +3 -0
  20. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/1.0 +3 -0
  21. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/2.0 +3 -0
  22. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/3.0 +3 -0
  23. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/4.0 +3 -0
  24. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/5.0 +3 -0
  25. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/6.0 +3 -0
  26. checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/7.0 +3 -0
  27. checkpoint_1007000/target.decoder.layers_21.pre_mlp_layer_norm.scale/0 +0 -0
  28. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray +1 -0
  29. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp12935395085792080493.~1663960300386769~ +1 -0
  30. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp1405165851374102366.~1663960300824635~ +1 -0
  31. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp15533635209282082156.~1663960300685855~ +1 -0
  32. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp3481497421862117506.~1663960300622157~ +1 -0
  33. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp5216179470711218956.~1663960300599429~ +1 -0
  34. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp6041758501264160899.~1663960300465960~ +1 -0
  35. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp9496050934012380735.~1663960300777385~ +1 -0
  36. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.0 +3 -0
  37. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.1 +3 -0
  38. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.2 +3 -0
  39. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.3 +3 -0
  40. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.4 +3 -0
  41. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.5 +3 -0
  42. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.6 +3 -0
  43. checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.7 +3 -0
  44. checkpoint_1007000/target.decoder.layers_8.pre_mlp_layer_norm.scale/.zarray +1 -0
  45. checkpoint_1007000/target.decoder.layers_8.pre_mlp_layer_norm.scale/0 +0 -0
  46. checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray +1 -0
  47. checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp1200407996780146747.~1663960300500697~ +1 -0
  48. checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp12015907105205464456.~1663960300468258~ +1 -0
  49. checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp17002564679374214580.~1663960300622628~ +1 -0
  50. checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp17609380261191980303.~1663960300551125~ +1 -0
.gitattributes CHANGED
@@ -30,3 +30,48 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
30
  *.zip filter=lfs diff=lfs merge=lfs -text
31
  *.zst filter=lfs diff=lfs merge=lfs -text
32
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
30
  *.zip filter=lfs diff=lfs merge=lfs -text
31
  *.zst filter=lfs diff=lfs merge=lfs -text
32
  *tfevents* filter=lfs diff=lfs merge=lfs -text
33
+ checkpoint_1007000/train_ds-003-of-004.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
34
+ checkpoint_1007000/train_ds-001-of-004.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
35
+ checkpoint_1007000/checkpoint filter=lfs diff=lfs merge=lfs -text
36
+ checkpoint_1007000/train_ds-000-of-004.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
37
+ checkpoint_1007000/train_ds-002-of-004.data-00000-of-00001 filter=lfs diff=lfs merge=lfs -text
38
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.1 filter=lfs diff=lfs merge=lfs -text
39
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.7 filter=lfs diff=lfs merge=lfs -text
40
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.0 filter=lfs diff=lfs merge=lfs -text
41
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.6 filter=lfs diff=lfs merge=lfs -text
42
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.5 filter=lfs diff=lfs merge=lfs -text
43
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.3 filter=lfs diff=lfs merge=lfs -text
44
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.2 filter=lfs diff=lfs merge=lfs -text
45
+ checkpoint_1007000/target.encoder.layers_22.attention.query.kernel/0.4 filter=lfs diff=lfs merge=lfs -text
46
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.1 filter=lfs diff=lfs merge=lfs -text
47
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.7 filter=lfs diff=lfs merge=lfs -text
48
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.0 filter=lfs diff=lfs merge=lfs -text
49
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.6 filter=lfs diff=lfs merge=lfs -text
50
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.5 filter=lfs diff=lfs merge=lfs -text
51
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.3 filter=lfs diff=lfs merge=lfs -text
52
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.2 filter=lfs diff=lfs merge=lfs -text
53
+ checkpoint_1007000/target.encoder.layers_3.mlp.wi_0.kernel/0.4 filter=lfs diff=lfs merge=lfs -text
54
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.1 filter=lfs diff=lfs merge=lfs -text
55
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.7 filter=lfs diff=lfs merge=lfs -text
56
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.0 filter=lfs diff=lfs merge=lfs -text
57
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.6 filter=lfs diff=lfs merge=lfs -text
58
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.5 filter=lfs diff=lfs merge=lfs -text
59
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.3 filter=lfs diff=lfs merge=lfs -text
60
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.2 filter=lfs diff=lfs merge=lfs -text
61
+ checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.4 filter=lfs diff=lfs merge=lfs -text
62
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/6.0 filter=lfs diff=lfs merge=lfs -text
63
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/1.0 filter=lfs diff=lfs merge=lfs -text
64
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/3.0 filter=lfs diff=lfs merge=lfs -text
65
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/0.0 filter=lfs diff=lfs merge=lfs -text
66
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/7.0 filter=lfs diff=lfs merge=lfs -text
67
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/5.0 filter=lfs diff=lfs merge=lfs -text
68
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/2.0 filter=lfs diff=lfs merge=lfs -text
69
+ checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/4.0 filter=lfs diff=lfs merge=lfs -text
70
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/6.0 filter=lfs diff=lfs merge=lfs -text
71
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/1.0 filter=lfs diff=lfs merge=lfs -text
72
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/3.0 filter=lfs diff=lfs merge=lfs -text
73
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/0.0 filter=lfs diff=lfs merge=lfs -text
74
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/7.0 filter=lfs diff=lfs merge=lfs -text
75
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/5.0 filter=lfs diff=lfs merge=lfs -text
76
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/2.0 filter=lfs diff=lfs merge=lfs -text
77
+ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/4.0 filter=lfs diff=lfs merge=lfs -text
checkpoint_1007000/checkpoint ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9dd585cc57ec6ce27ec16da81a66e38b816d4c341a807471d5618fbe798d9873
3
+ size 19920805
checkpoint_1007000/state.param_states.decoder.layers_14.pre_mlp_layer_norm.scale.v/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096],"zarr_format":2}
checkpoint_1007000/state.param_states.decoder.layers_14.pre_mlp_layer_norm.scale.v/0 ADDED
Binary file (14.9 kB). View file
 
checkpoint_1007000/state.param_states.decoder.layers_19.pre_self_attention_layer_norm.scale.v/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096],"zarr_format":2}
checkpoint_1007000/state.param_states.decoder.layers_19.pre_self_attention_layer_norm.scale.v/0 ADDED
Binary file (14.7 kB). View file
 
checkpoint_1007000/state.param_states.decoder.layers_5.pre_mlp_layer_norm.scale.v/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096],"zarr_format":2}
checkpoint_1007000/state.param_states.decoder.layers_5.pre_mlp_layer_norm.scale.v/0 ADDED
Binary file (14.8 kB). View file
 
checkpoint_1007000/state.param_states.decoder.layers_7.pre_mlp_layer_norm.scale.v/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096],"zarr_format":2}
checkpoint_1007000/state.param_states.decoder.layers_7.pre_mlp_layer_norm.scale.v/0 ADDED
Binary file (14.9 kB). View file
 
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp11477047902703534671.~1663960300386635~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp2065349081025687113.~1663960300549585~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp444933188802333506.~1663960300873156~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp4798233591127119695.~1663960300685739~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp5336578133756373591.~1663960300505112~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp8112336193798835172.~1663960300777247~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/.zarray.__tmp9467005472495202151.~1663960300500527~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[512,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/0.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ee34b58a8681b7cb7f4ec159cf464705d8086b3b2059dbf4144b5585116403f
3
+ size 7802006
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/1.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2c9ef70da8284b3fe44827008e242135e346e4d665eab4b2ed12e9a45601ead
3
+ size 7802592
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/2.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ff8566a0ba1961daf34583375b959405d0925084991d1c90f1e8040e14bac2c7
3
+ size 7802819
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/3.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b17852527f669e14cb2e670508c97a63e971bff80cb606b73859314e63e8ca72
3
+ size 7802754
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/4.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:da0a505e923ec14491265ff9bdcccea7a7a3920fdc58c145dbf0ae779eb42cd2
3
+ size 7803020
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/5.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6408d9617975044a597a02378afe7c369461c459fc28c6e0a68030bf0a821bd
3
+ size 7803432
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/6.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40f81bac422c2370bbe2391e8dbb0e1b9fdb941e66f22f18f28a1c14efd98572
3
+ size 7803137
checkpoint_1007000/target.decoder.layers_21.encoder_decoder_attention.out.kernel/7.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6bfb9b1010c745946085ffab2e3190db689fb2ec66d476caa7829c9a49d79609
3
+ size 7803379
checkpoint_1007000/target.decoder.layers_21.pre_mlp_layer_norm.scale/0 ADDED
Binary file (13.8 kB). View file
 
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp12935395085792080493.~1663960300386769~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp1405165851374102366.~1663960300824635~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp15533635209282082156.~1663960300685855~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp3481497421862117506.~1663960300622157~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp5216179470711218956.~1663960300599429~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp6041758501264160899.~1663960300465960~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/.zarray.__tmp9496050934012380735.~1663960300777385~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096,512],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096,4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d73969141122fac82f0d6e8b8fa791cab402cbce5b70ff038dd3eb3f8834052
3
+ size 7804879
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:089ceba0069ecba3a352c58836b74323f5323e2c74f9ef37ebb0a0207d4bfb82
3
+ size 7813408
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a2f5ade1e327cf6fc8ab6e6eca5360077f15936a8d75c0fa4acb6e122d218c1
3
+ size 7808993
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.3 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:71bfaac90516533307508762db6ea1368bc8341d904312266a328b95c5d40b91
3
+ size 7813166
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4972058abcfa2bb9f8ce007d8a32fec0a35e8fb2c2a2efcf626c85d483e104a7
3
+ size 7805901
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f73685f6d9d3c30a5fe8d8dea7cd3427256254a51a27883ce94e15139e88751
3
+ size 7805611
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.6 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f56ecece74b48db366f38b5e6eda6fefbc6a9b7c77219e1dcbf2ce8810347df
3
+ size 7804647
checkpoint_1007000/target.decoder.layers_21.self_attention.query.kernel/0.7 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf6a3c700f52a9bb7b29838289139eed3d36902fff13f437fd1a0be65dd708fe
3
+ size 7804297
checkpoint_1007000/target.decoder.layers_8.pre_mlp_layer_norm.scale/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4096],"zarr_format":2}
checkpoint_1007000/target.decoder.layers_8.pre_mlp_layer_norm.scale/0 ADDED
Binary file (14.9 kB). View file
 
checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[1280,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[10240,4096],"zarr_format":2}
checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp1200407996780146747.~1663960300500697~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[1280,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[10240,4096],"zarr_format":2}
checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp12015907105205464456.~1663960300468258~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[1280,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[10240,4096],"zarr_format":2}
checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp17002564679374214580.~1663960300622628~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[1280,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[10240,4096],"zarr_format":2}
checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/.zarray.__tmp17609380261191980303.~1663960300551125~ ADDED
@@ -0,0 +1 @@
 
 
1
+ {"chunks":[1280,4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[10240,4096],"zarr_format":2}