Upload HiDreamImagePipeline

README.md

@@ -1,3 +1,6 @@
+---
+library_name: diffusers
+---
 created using the following code based on https://github.com/huggingface/diffusers/blob/main/tests/pipelines/hidream/test_pipeline_hidream.py
 
 ```python

model_index.json

@@ -1,6 +1,6 @@
 {
   "_class_name": "HiDreamImagePipeline",
-  "_diffusers_version": "0.33.0.dev0",
+  "_diffusers_version": "0.34.0.dev0",
   "scheduler": [
     "diffusers",
     "FlowMatchEulerDiscreteScheduler"
@@ -31,7 +31,7 @@
   ],
   "tokenizer_3": [
     "transformers",
-    "T5TokenizerFast"
+    "T5Tokenizer"
   ],
   "tokenizer_4": [
     "transformers",

scheduler/scheduler_config.json

@@ -1,6 +1,6 @@
 {
   "_class_name": "FlowMatchEulerDiscreteScheduler",
-  "_diffusers_version": "0.33.0.dev0",
+  "_diffusers_version": "0.34.0.dev0",
   "base_image_seq_len": 256,
   "base_shift": 0.5,
   "invert_sigmas": false,

text_encoder/config.json

@@ -18,6 +18,6 @@
   "pad_token_id": 1,
   "projection_dim": 32,
   "torch_dtype": "float32",
-  "transformers_version": "4.49.0",
+  "transformers_version": "4.51.3",
   "vocab_size": 1000
 }

text_encoder_2/config.json

@@ -18,6 +18,6 @@
   "pad_token_id": 1,
   "projection_dim": 32,
   "torch_dtype": "float32",
-  "transformers_version": "4.49.0",
+  "transformers_version": "4.51.3",
   "vocab_size": 1000
 }

text_encoder_3/config.json

@@ -1,5 +1,4 @@
 {
-  "_name_or_path": "hf-internal-testing/tiny-random-t5",
   "architectures": [
     "T5EncoderModel"
   ],
@@ -13,7 +12,6 @@
   "dropout_rate": 0.1,
   "eos_token_id": 1,
   "feed_forward_proj": "relu",
-  "gradient_checkpointing": false,
   "initializer_factor": 0.002,
   "is_encoder_decoder": true,
   "is_gated_act": false,
@@ -26,7 +24,7 @@
   "relative_attention_max_distance": 128,
   "relative_attention_num_buckets": 8,
   "torch_dtype": "float32",
-  "transformers_version": "4.49.0",
+  "transformers_version": "4.51.3",
   "use_cache": true,
-  "vocab_size": 1103
+  "vocab_size": 32100
 }

text_encoder_3/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a3ff1d3d80459481c9847d35de9a618835224139a4de455b40fcfe6f8544193b
-size 276696
+oid sha256:ffa27ba9ee9488f85f598191955b5bce34db1a0fe37bb34ec27456298ccc204c
+size 4244320

text_encoder_4/config.json

@@ -1,5 +1,4 @@
 {
-  "_name_or_path": "hf-internal-testing/tiny-random-LlamaForCausalLM",
   "architectures": [
     "LlamaForCausalLM"
   ],
@@ -25,7 +24,7 @@
   "rope_theta": 10000.0,
   "tie_word_embeddings": false,
   "torch_dtype": "float32",
-  "transformers_version": "4.49.0",
+  "transformers_version": "4.51.3",
   "use_cache": true,
   "vocab_size": 32000
 }

text_encoder_4/generation_config.json

@@ -3,5 +3,5 @@
   "bos_token_id": 0,
   "eos_token_id": 1,
   "pad_token_id": 1,
-  "transformers_version": "4.49.0"
+  "transformers_version": "4.51.3"
 }

tokenizer_3/added_tokens.json

@@ -0,0 +1,102 @@
+{
+  "<extra_id_0>": 32099,
+  "<extra_id_10>": 32089,
+  "<extra_id_11>": 32088,
+  "<extra_id_12>": 32087,
+  "<extra_id_13>": 32086,
+  "<extra_id_14>": 32085,
+  "<extra_id_15>": 32084,
+  "<extra_id_16>": 32083,
+  "<extra_id_17>": 32082,
+  "<extra_id_18>": 32081,
+  "<extra_id_19>": 32080,
+  "<extra_id_1>": 32098,
+  "<extra_id_20>": 32079,
+  "<extra_id_21>": 32078,
+  "<extra_id_22>": 32077,
+  "<extra_id_23>": 32076,
+  "<extra_id_24>": 32075,
+  "<extra_id_25>": 32074,
+  "<extra_id_26>": 32073,
+  "<extra_id_27>": 32072,
+  "<extra_id_28>": 32071,
+  "<extra_id_29>": 32070,
+  "<extra_id_2>": 32097,
+  "<extra_id_30>": 32069,
+  "<extra_id_31>": 32068,
+  "<extra_id_32>": 32067,
+  "<extra_id_33>": 32066,
+  "<extra_id_34>": 32065,
+  "<extra_id_35>": 32064,
+  "<extra_id_36>": 32063,
+  "<extra_id_37>": 32062,
+  "<extra_id_38>": 32061,
+  "<extra_id_39>": 32060,
+  "<extra_id_3>": 32096,
+  "<extra_id_40>": 32059,
+  "<extra_id_41>": 32058,
+  "<extra_id_42>": 32057,
+  "<extra_id_43>": 32056,
+  "<extra_id_44>": 32055,
+  "<extra_id_45>": 32054,
+  "<extra_id_46>": 32053,
+  "<extra_id_47>": 32052,
+  "<extra_id_48>": 32051,
+  "<extra_id_49>": 32050,
+  "<extra_id_4>": 32095,
+  "<extra_id_50>": 32049,
+  "<extra_id_51>": 32048,
+  "<extra_id_52>": 32047,
+  "<extra_id_53>": 32046,
+  "<extra_id_54>": 32045,
+  "<extra_id_55>": 32044,
+  "<extra_id_56>": 32043,
+  "<extra_id_57>": 32042,
+  "<extra_id_58>": 32041,
+  "<extra_id_59>": 32040,
+  "<extra_id_5>": 32094,
+  "<extra_id_60>": 32039,
+  "<extra_id_61>": 32038,
+  "<extra_id_62>": 32037,
+  "<extra_id_63>": 32036,
+  "<extra_id_64>": 32035,
+  "<extra_id_65>": 32034,
+  "<extra_id_66>": 32033,
+  "<extra_id_67>": 32032,
+  "<extra_id_68>": 32031,
+  "<extra_id_69>": 32030,
+  "<extra_id_6>": 32093,
+  "<extra_id_70>": 32029,
+  "<extra_id_71>": 32028,
+  "<extra_id_72>": 32027,
+  "<extra_id_73>": 32026,
+  "<extra_id_74>": 32025,
+  "<extra_id_75>": 32024,
+  "<extra_id_76>": 32023,
+  "<extra_id_77>": 32022,
+  "<extra_id_78>": 32021,
+  "<extra_id_79>": 32020,
+  "<extra_id_7>": 32092,
+  "<extra_id_80>": 32019,
+  "<extra_id_81>": 32018,
+  "<extra_id_82>": 32017,
+  "<extra_id_83>": 32016,
+  "<extra_id_84>": 32015,
+  "<extra_id_85>": 32014,
+  "<extra_id_86>": 32013,
+  "<extra_id_87>": 32012,
+  "<extra_id_88>": 32011,
+  "<extra_id_89>": 32010,
+  "<extra_id_8>": 32091,
+  "<extra_id_90>": 32009,
+  "<extra_id_91>": 32008,
+  "<extra_id_92>": 32007,
+  "<extra_id_93>": 32006,
+  "<extra_id_94>": 32005,
+  "<extra_id_95>": 32004,
+  "<extra_id_96>": 32003,
+  "<extra_id_97>": 32002,
+  "<extra_id_98>": 32001,
+  "<extra_id_99>": 32000,
+  "<extra_id_9>": 32090
+}

tokenizer_3/spiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d60acb128cf7b7f2536e8f38a5b18a05535c9e14c7a355904270e15b0945ea86
+size 791656

tokenizer_3/tokenizer_config.json

@@ -1,5 +1,5 @@
 {
-  "add_prefix_space": null,
+  "add_prefix_space": true,
   "added_tokens_decoder": {
     "0": {
       "content": "<pad>",
@@ -25,7 +25,7 @@
       "single_word": false,
       "special": true
     },
-    "3": {
+    "32000": {
       "content": "<extra_id_99>",
       "lstrip": false,
       "normalized": false,
@@ -33,7 +33,7 @@
       "single_word": false,
       "special": true
     },
-    "4": {
+    "32001": {
       "content": "<extra_id_98>",
       "lstrip": false,
       "normalized": false,
@@ -41,7 +41,7 @@
       "single_word": false,
       "special": true
     },
-    "5": {
+    "32002": {
       "content": "<extra_id_97>",
       "lstrip": false,
       "normalized": false,
@@ -49,7 +49,7 @@
       "single_word": false,
       "special": true
     },
-    "6": {
+    "32003": {
       "content": "<extra_id_96>",
       "lstrip": false,
       "normalized": false,
@@ -57,7 +57,7 @@
       "single_word": false,
       "special": true
     },
-    "7": {
+    "32004": {
       "content": "<extra_id_95>",
       "lstrip": false,
       "normalized": false,
@@ -65,7 +65,7 @@
       "single_word": false,
       "special": true
     },
-    "8": {
+    "32005": {
       "content": "<extra_id_94>",
       "lstrip": false,
       "normalized": false,
@@ -73,7 +73,7 @@
       "single_word": false,
       "special": true
     },
-    "9": {
+    "32006": {
       "content": "<extra_id_93>",
       "lstrip": false,
       "normalized": false,
@@ -81,7 +81,7 @@
       "single_word": false,
       "special": true
     },
-    "10": {
+    "32007": {
       "content": "<extra_id_92>",
       "lstrip": false,
       "normalized": false,
@@ -89,7 +89,7 @@
       "single_word": false,
       "special": true
     },
-    "11": {
+    "32008": {
       "content": "<extra_id_91>",
       "lstrip": false,
       "normalized": false,
@@ -97,7 +97,7 @@
       "single_word": false,
       "special": true
     },
-    "12": {
+    "32009": {
       "content": "<extra_id_90>",
       "lstrip": false,
       "normalized": false,
@@ -105,7 +105,7 @@
       "single_word": false,
       "special": true
     },
-    "13": {
+    "32010": {
       "content": "<extra_id_89>",
       "lstrip": false,
       "normalized": false,
@@ -113,7 +113,7 @@
       "single_word": false,
       "special": true
     },
-    "14": {
+    "32011": {
       "content": "<extra_id_88>",
       "lstrip": false,
       "normalized": false,
@@ -121,7 +121,7 @@
       "single_word": false,
       "special": true
     },
-    "15": {
+    "32012": {
       "content": "<extra_id_87>",
       "lstrip": false,
       "normalized": false,
@@ -129,7 +129,7 @@
       "single_word": false,
       "special": true
     },
-    "16": {
+    "32013": {
       "content": "<extra_id_86>",
       "lstrip": false,
       "normalized": false,
@@ -137,7 +137,7 @@
       "single_word": false,
       "special": true
     },
-    "17": {
+    "32014": {
       "content": "<extra_id_85>",
       "lstrip": false,
       "normalized": false,
@@ -145,7 +145,7 @@
       "single_word": false,
       "special": true
     },
-    "18": {
+    "32015": {
       "content": "<extra_id_84>",
       "lstrip": false,
       "normalized": false,
@@ -153,7 +153,7 @@
       "single_word": false,
       "special": true
     },
-    "19": {
+    "32016": {
       "content": "<extra_id_83>",
       "lstrip": false,
       "normalized": false,
@@ -161,7 +161,7 @@
       "single_word": false,
       "special": true
     },
-    "20": {
+    "32017": {
       "content": "<extra_id_82>",
       "lstrip": false,
       "normalized": false,
@@ -169,7 +169,7 @@
       "single_word": false,
       "special": true
     },
-    "21": {
+    "32018": {
       "content": "<extra_id_81>",
       "lstrip": false,
       "normalized": false,
@@ -177,7 +177,7 @@
       "single_word": false,
       "special": true
     },
-    "22": {
+    "32019": {
       "content": "<extra_id_80>",
       "lstrip": false,
       "normalized": false,
@@ -185,7 +185,7 @@
       "single_word": false,
       "special": true
     },
-    "23": {
+    "32020": {
       "content": "<extra_id_79>",
       "lstrip": false,
       "normalized": false,
@@ -193,7 +193,7 @@
       "single_word": false,
       "special": true
     },
-    "24": {
+    "32021": {
       "content": "<extra_id_78>",
       "lstrip": false,
       "normalized": false,
@@ -201,7 +201,7 @@
       "single_word": false,
       "special": true
     },
-    "25": {
+    "32022": {
       "content": "<extra_id_77>",
       "lstrip": false,
       "normalized": false,
@@ -209,7 +209,7 @@
       "single_word": false,
       "special": true
     },
-    "26": {
+    "32023": {
       "content": "<extra_id_76>",
       "lstrip": false,
       "normalized": false,
@@ -217,7 +217,7 @@
       "single_word": false,
       "special": true
     },
-    "27": {
+    "32024": {
       "content": "<extra_id_75>",
       "lstrip": false,
       "normalized": false,
@@ -225,7 +225,7 @@
       "single_word": false,
       "special": true
     },
-    "28": {
+    "32025": {
       "content": "<extra_id_74>",
       "lstrip": false,
       "normalized": false,
@@ -233,7 +233,7 @@
       "single_word": false,
       "special": true
     },
-    "29": {
+    "32026": {
       "content": "<extra_id_73>",
       "lstrip": false,
       "normalized": false,
@@ -241,7 +241,7 @@
       "single_word": false,
       "special": true
     },
-    "30": {
+    "32027": {
       "content": "<extra_id_72>",
       "lstrip": false,
       "normalized": false,
@@ -249,7 +249,7 @@
       "single_word": false,
       "special": true
     },
-    "31": {
+    "32028": {
       "content": "<extra_id_71>",
       "lstrip": false,
       "normalized": false,
@@ -257,7 +257,7 @@
       "single_word": false,
       "special": true
     },
-    "32": {
+    "32029": {
       "content": "<extra_id_70>",
       "lstrip": false,
       "normalized": false,
@@ -265,7 +265,7 @@
       "single_word": false,
       "special": true
     },
-    "33": {
+    "32030": {
       "content": "<extra_id_69>",
       "lstrip": false,
       "normalized": false,
@@ -273,7 +273,7 @@
       "single_word": false,
       "special": true
     },
-    "34": {
+    "32031": {
       "content": "<extra_id_68>",
       "lstrip": false,
       "normalized": false,
@@ -281,7 +281,7 @@
       "single_word": false,
       "special": true
     },
-    "35": {
+    "32032": {
       "content": "<extra_id_67>",
       "lstrip": false,
       "normalized": false,
@@ -289,7 +289,7 @@
       "single_word": false,
       "special": true
     },
-    "36": {
+    "32033": {
       "content": "<extra_id_66>",
       "lstrip": false,
       "normalized": false,
@@ -297,7 +297,7 @@
       "single_word": false,
       "special": true
     },
-    "37": {
+    "32034": {
       "content": "<extra_id_65>",
       "lstrip": false,
       "normalized": false,
@@ -305,7 +305,7 @@
       "single_word": false,
       "special": true
     },
-    "38": {
+    "32035": {
       "content": "<extra_id_64>",
       "lstrip": false,
       "normalized": false,
@@ -313,7 +313,7 @@
       "single_word": false,
       "special": true
     },
-    "39": {
+    "32036": {
       "content": "<extra_id_63>",
       "lstrip": false,
       "normalized": false,
@@ -321,7 +321,7 @@
       "single_word": false,
       "special": true
     },
-    "40": {
+    "32037": {
       "content": "<extra_id_62>",
       "lstrip": false,
       "normalized": false,
@@ -329,7 +329,7 @@
       "single_word": false,
       "special": true
     },
-    "41": {
+    "32038": {
       "content": "<extra_id_61>",
       "lstrip": false,
       "normalized": false,
@@ -337,7 +337,7 @@
       "single_word": false,
       "special": true
     },
-    "42": {
+    "32039": {
       "content": "<extra_id_60>",
       "lstrip": false,
       "normalized": false,
@@ -345,7 +345,7 @@
       "single_word": false,
       "special": true
     },
-    "43": {
+    "32040": {
       "content": "<extra_id_59>",
       "lstrip": false,
       "normalized": false,
@@ -353,7 +353,7 @@
       "single_word": false,
       "special": true
     },
-    "44": {
+    "32041": {
       "content": "<extra_id_58>",
       "lstrip": false,
       "normalized": false,
@@ -361,7 +361,7 @@
       "single_word": false,
       "special": true
     },
-    "45": {
+    "32042": {
       "content": "<extra_id_57>",
       "lstrip": false,
       "normalized": false,
@@ -369,7 +369,7 @@
       "single_word": false,
       "special": true
     },
-    "46": {
+    "32043": {
       "content": "<extra_id_56>",
       "lstrip": false,
       "normalized": false,
@@ -377,7 +377,7 @@
       "single_word": false,
       "special": true
     },
-    "47": {
+    "32044": {
       "content": "<extra_id_55>",
       "lstrip": false,
       "normalized": false,
@@ -385,7 +385,7 @@
       "single_word": false,
       "special": true
     },
-    "48": {
+    "32045": {
       "content": "<extra_id_54>",
       "lstrip": false,
       "normalized": false,
@@ -393,7 +393,7 @@
       "single_word": false,
       "special": true
     },
-    "49": {
+    "32046": {
       "content": "<extra_id_53>",
       "lstrip": false,
       "normalized": false,
@@ -401,7 +401,7 @@
       "single_word": false,
       "special": true
     },
-    "50": {
+    "32047": {
       "content": "<extra_id_52>",
       "lstrip": false,
       "normalized": false,
@@ -409,7 +409,7 @@
       "single_word": false,
       "special": true
     },
-    "51": {
+    "32048": {
       "content": "<extra_id_51>",
       "lstrip": false,
       "normalized": false,
@@ -417,7 +417,7 @@
       "single_word": false,
       "special": true
     },
-    "52": {
+    "32049": {
       "content": "<extra_id_50>",
       "lstrip": false,
       "normalized": false,
@@ -425,7 +425,7 @@
       "single_word": false,
       "special": true
     },
-    "53": {
+    "32050": {
       "content": "<extra_id_49>",
       "lstrip": false,
       "normalized": false,
@@ -433,7 +433,7 @@
       "single_word": false,
       "special": true
     },
-    "54": {
+    "32051": {
       "content": "<extra_id_48>",
       "lstrip": false,
       "normalized": false,
@@ -441,7 +441,7 @@
       "single_word": false,
       "special": true
     },
-    "55": {
+    "32052": {
       "content": "<extra_id_47>",
       "lstrip": false,
       "normalized": false,
@@ -449,7 +449,7 @@
       "single_word": false,
       "special": true
     },
-    "56": {
+    "32053": {
       "content": "<extra_id_46>",
       "lstrip": false,
       "normalized": false,
@@ -457,7 +457,7 @@
       "single_word": false,
       "special": true
     },
-    "57": {
+    "32054": {
       "content": "<extra_id_45>",
       "lstrip": false,
       "normalized": false,
@@ -465,7 +465,7 @@
       "single_word": false,
       "special": true
     },
-    "58": {
+    "32055": {
       "content": "<extra_id_44>",
       "lstrip": false,
       "normalized": false,
@@ -473,7 +473,7 @@
       "single_word": false,
       "special": true
     },
-    "59": {
+    "32056": {
       "content": "<extra_id_43>",
       "lstrip": false,
       "normalized": false,
@@ -481,7 +481,7 @@
       "single_word": false,
       "special": true
     },
-    "60": {
+    "32057": {
       "content": "<extra_id_42>",
       "lstrip": false,
       "normalized": false,
@@ -489,7 +489,7 @@
       "single_word": false,
       "special": true
     },
-    "61": {
+    "32058": {
       "content": "<extra_id_41>",
       "lstrip": false,
       "normalized": false,
@@ -497,7 +497,7 @@
       "single_word": false,
       "special": true
     },
-    "62": {
+    "32059": {
       "content": "<extra_id_40>",
       "lstrip": false,
       "normalized": false,
@@ -505,7 +505,7 @@
       "single_word": false,
       "special": true
     },
-    "63": {
+    "32060": {
       "content": "<extra_id_39>",
       "lstrip": false,
       "normalized": false,
@@ -513,7 +513,7 @@
       "single_word": false,
       "special": true
     },
-    "64": {
+    "32061": {
       "content": "<extra_id_38>",
       "lstrip": false,
       "normalized": false,
@@ -521,7 +521,7 @@
       "single_word": false,
       "special": true
     },
-    "65": {
+    "32062": {
       "content": "<extra_id_37>",
       "lstrip": false,
       "normalized": false,
@@ -529,7 +529,7 @@
       "single_word": false,
       "special": true
     },
-    "66": {
+    "32063": {
       "content": "<extra_id_36>",
       "lstrip": false,
       "normalized": false,
@@ -537,7 +537,7 @@
       "single_word": false,
       "special": true
     },
-    "67": {
+    "32064": {
       "content": "<extra_id_35>",
       "lstrip": false,
       "normalized": false,
@@ -545,7 +545,7 @@
       "single_word": false,
       "special": true
     },
-    "68": {
+    "32065": {
       "content": "<extra_id_34>",
       "lstrip": false,
       "normalized": false,
@@ -553,7 +553,7 @@
       "single_word": false,
       "special": true
     },
-    "69": {
+    "32066": {
       "content": "<extra_id_33>",
       "lstrip": false,
       "normalized": false,
@@ -561,7 +561,7 @@
       "single_word": false,
       "special": true
     },
-    "70": {
+    "32067": {
       "content": "<extra_id_32>",
       "lstrip": false,
       "normalized": false,
@@ -569,7 +569,7 @@
       "single_word": false,
       "special": true
     },
-    "71": {
+    "32068": {
       "content": "<extra_id_31>",
       "lstrip": false,
       "normalized": false,
@@ -577,7 +577,7 @@
       "single_word": false,
       "special": true
     },
-    "72": {
+    "32069": {
       "content": "<extra_id_30>",
       "lstrip": false,
       "normalized": false,
@@ -585,7 +585,7 @@
       "single_word": false,
       "special": true
     },
-    "73": {
+    "32070": {
       "content": "<extra_id_29>",
       "lstrip": false,
       "normalized": false,
@@ -593,7 +593,7 @@
       "single_word": false,
       "special": true
     },
-    "74": {
+    "32071": {
       "content": "<extra_id_28>",
       "lstrip": false,
       "normalized": false,
@@ -601,7 +601,7 @@
       "single_word": false,
       "special": true
     },
-    "75": {
+    "32072": {
       "content": "<extra_id_27>",
       "lstrip": false,
       "normalized": false,
@@ -609,7 +609,7 @@
       "single_word": false,
       "special": true
     },
-    "76": {
+    "32073": {
       "content": "<extra_id_26>",
       "lstrip": false,
       "normalized": false,
@@ -617,7 +617,7 @@
       "single_word": false,
       "special": true
     },
-    "77": {
+    "32074": {
       "content": "<extra_id_25>",
       "lstrip": false,
       "normalized": false,
@@ -625,7 +625,7 @@
       "single_word": false,
       "special": true
     },
-    "78": {
+    "32075": {
       "content": "<extra_id_24>",
       "lstrip": false,
       "normalized": false,
@@ -633,7 +633,7 @@
       "single_word": false,
       "special": true
     },
-    "79": {
+    "32076": {
       "content": "<extra_id_23>",
       "lstrip": false,
       "normalized": false,
@@ -641,7 +641,7 @@
       "single_word": false,
       "special": true
     },
-    "80": {
+    "32077": {
       "content": "<extra_id_22>",
       "lstrip": false,
       "normalized": false,
@@ -649,7 +649,7 @@
       "single_word": false,
       "special": true
     },
-    "81": {
+    "32078": {
       "content": "<extra_id_21>",
       "lstrip": false,
       "normalized": false,
@@ -657,7 +657,7 @@
       "single_word": false,
       "special": true
     },
-    "82": {
+    "32079": {
       "content": "<extra_id_20>",
       "lstrip": false,
       "normalized": false,
@@ -665,7 +665,7 @@
       "single_word": false,
       "special": true
     },
-    "83": {
+    "32080": {
       "content": "<extra_id_19>",
       "lstrip": false,
       "normalized": false,
@@ -673,7 +673,7 @@
       "single_word": false,
       "special": true
     },
-    "84": {
+    "32081": {
       "content": "<extra_id_18>",
       "lstrip": false,
       "normalized": false,
@@ -681,7 +681,7 @@
       "single_word": false,
       "special": true
     },
-    "85": {
+    "32082": {
       "content": "<extra_id_17>",
       "lstrip": false,
       "normalized": false,
@@ -689,7 +689,7 @@
       "single_word": false,
       "special": true
     },
-    "86": {
+    "32083": {
       "content": "<extra_id_16>",
       "lstrip": false,
       "normalized": false,
@@ -697,7 +697,7 @@
       "single_word": false,
       "special": true
     },
-    "87": {
+    "32084": {
       "content": "<extra_id_15>",
       "lstrip": false,
       "normalized": false,
@@ -705,7 +705,7 @@
       "single_word": false,
       "special": true
     },
-    "88": {
+    "32085": {
       "content": "<extra_id_14>",
       "lstrip": false,
       "normalized": false,
@@ -713,7 +713,7 @@
       "single_word": false,
       "special": true
     },
-    "89": {
+    "32086": {
       "content": "<extra_id_13>",
       "lstrip": false,
       "normalized": false,
@@ -721,7 +721,7 @@
       "single_word": false,
       "special": true
     },
-    "90": {
+    "32087": {
       "content": "<extra_id_12>",
       "lstrip": false,
       "normalized": false,
@@ -729,7 +729,7 @@
       "single_word": false,
       "special": true
     },
-    "91": {
+    "32088": {
       "content": "<extra_id_11>",
       "lstrip": false,
       "normalized": false,
@@ -737,7 +737,7 @@
       "single_word": false,
       "special": true
     },
-    "92": {
+    "32089": {
       "content": "<extra_id_10>",
       "lstrip": false,
       "normalized": false,
@@ -745,7 +745,7 @@
       "single_word": false,
       "special": true
     },
-    "93": {
+    "32090": {
       "content": "<extra_id_9>",
       "lstrip": false,
       "normalized": false,
@@ -753,7 +753,7 @@
       "single_word": false,
       "special": true
     },
-    "94": {
+    "32091": {
       "content": "<extra_id_8>",
       "lstrip": false,
       "normalized": false,
@@ -761,7 +761,7 @@
       "single_word": false,
       "special": true
     },
-    "95": {
+    "32092": {
       "content": "<extra_id_7>",
       "lstrip": false,
       "normalized": false,
@@ -769,7 +769,7 @@
       "single_word": false,
       "special": true
     },
-    "96": {
+    "32093": {
       "content": "<extra_id_6>",
       "lstrip": false,
       "normalized": false,
@@ -777,7 +777,7 @@
       "single_word": false,
       "special": true
     },
-    "97": {
+    "32094": {
       "content": "<extra_id_5>",
       "lstrip": false,
       "normalized": false,
@@ -785,7 +785,7 @@
       "single_word": false,
       "special": true
     },
-    "98": {
+    "32095": {
       "content": "<extra_id_4>",
       "lstrip": false,
       "normalized": false,
@@ -793,7 +793,7 @@
       "single_word": false,
       "special": true
     },
-    "99": {
+    "32096": {
       "content": "<extra_id_3>",
       "lstrip": false,
       "normalized": false,
@@ -801,7 +801,7 @@
       "single_word": false,
       "special": true
     },
-    "100": {
+    "32097": {
       "content": "<extra_id_2>",
       "lstrip": false,
       "normalized": false,
@@ -809,7 +809,7 @@
       "single_word": false,
       "special": true
     },
-    "101": {
+    "32098": {
       "content": "<extra_id_1>",
       "lstrip": false,
       "normalized": false,
@@ -817,7 +817,7 @@
       "single_word": false,
       "special": true
     },
-    "102": {
+    "32099": {
       "content": "<extra_id_0>",
       "lstrip": false,
       "normalized": false,
@@ -928,12 +928,14 @@
     "<extra_id_98>",
     "<extra_id_99>"
   ],
-  "clean_up_tokenization_spaces": false,
+  "clean_up_tokenization_spaces": true,
   "eos_token": "</s>",
   "extra_ids": 100,
   "extra_special_tokens": {},
-  "model_max_length": 1024,
+  "legacy": true,
+  "model_max_length": 512,
   "pad_token": "<pad>",
-  "tokenizer_class": "T5TokenizerFast",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "T5Tokenizer",
   "unk_token": "<unk>"
 }

transformer/config.json

@@ -1,6 +1,6 @@
 {
   "_class_name": "HiDreamImageTransformer2DModel",
-  "_diffusers_version": "0.33.0.dev0",
+  "_diffusers_version": "0.34.0.dev0",
   "attention_head_dim": 8,
   "axes_dims_rope": [
     4,
@@ -11,6 +11,7 @@
     32,
     16
   ],
+  "force_inference_output": false,
   "in_channels": 4,
   "llama_layers": [
     0,

vae/config.json

@@ -1,6 +1,6 @@
 {
   "_class_name": "AutoencoderKL",
-  "_diffusers_version": "0.33.0.dev0",
+  "_diffusers_version": "0.34.0.dev0",
   "act_fn": "silu",
   "block_out_channels": [
     64