Distillation test. lively-planet-17 (based on t5-efficient-tiny) distilled from trainer azure-frost-334 (baed on flan-t5-base)

config.json

@@ -1,61 +1,30 @@
 {
-  "_name_or_path": "eager-plant-323",
+  "_name_or_path": "lively-planet-17",
   "architectures": [
     "T5ForConditionalGeneration"
   ],
   "classifier_dropout": 0.0,
   "d_ff": 1024,
   "d_kv": 64,
-  "d_model": 512,
+  "d_model": 256,
   "decoder_start_token_id": 0,
-  "dense_act_fn": "gelu_new",
+  "dense_act_fn": "relu",
   "dropout_rate": 0.1,
   "eos_token_id": 1,
-  "feed_forward_proj": "gated-gelu",
+  "feed_forward_proj": "relu",
   "initializer_factor": 1.0,
   "is_encoder_decoder": true,
-  "is_gated_act": true,
+  "is_gated_act": false,
   "layer_norm_epsilon": 1e-06,
   "model_type": "t5",
   "n_positions": 512,
-  "num_decoder_layers": 8,
-  "num_heads": 6,
-  "num_layers": 8,
-  "output_past": true,
+  "num_decoder_layers": 4,
+  "num_heads": 4,
+  "num_layers": 4,
   "pad_token_id": 0,
   "relative_attention_max_distance": 128,
   "relative_attention_num_buckets": 32,
-  "task_specific_params": {
-    "summarization": {
-      "early_stopping": true,
-      "length_penalty": 2.0,
-      "max_length": 200,
-      "min_length": 30,
-      "no_repeat_ngram_size": 3,
-      "num_beams": 4,
-      "prefix": "summarize: "
-    },
-    "translation_en_to_de": {
-      "early_stopping": true,
-      "max_length": 300,
-      "num_beams": 4,
-      "prefix": "translate English to German: "
-    },
-    "translation_en_to_fr": {
-      "early_stopping": true,
-      "max_length": 300,
-      "num_beams": 4,
-      "prefix": "translate English to French: "
-    },
-    "translation_en_to_ro": {
-      "early_stopping": true,
-      "max_length": 300,
-      "num_beams": 4,
-      "prefix": "translate English to Romanian: "
-    }
-  },
-  "tie_word_embeddings": false,
   "transformers_version": "4.43.4",
-  "use_cache": true,
-  "vocab_size": 32128
+  "use_cache": false,
+  "vocab_size": 32100
 }

onnx/decoder_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6ccfd3430c89baf346caf4a250faa4da0c8b55e68a01e8272d887b18982b808f
-size 232559765
+oid sha256:3c25b421f4eb3b3e76288a7a8aca6f2eeaee00b0b357bcbc7ae058097ae67ddd
+size 49779670

onnx/decoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3f15c7fc318265c50cd49f0d8a1e98f355870274c32d1358c83f2b8ad99bb552
-size 58868682
+oid sha256:e272d12349b2cf1e99153afdd95b26b4212fe9a1e20894ddfeb0a04d90aa69e4
+size 45530444

onnx/encoder_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4d92762de4cd38898dd6c4004facf6148b55e224fd847cab65de9d4fa31fe0cb
-size 141467235
+oid sha256:ffc647d3cd898160c46aac53d6f79630467a34301fdae7ab653fc6b70ef8eba0
+size 45524751

onnx/encoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a1afa42a0bf17ec78ee965342836a1bb41f8ecc391edddf527812183df9c83fa
-size 35726832
+oid sha256:c7dcfd756fe8768b136b66e9d535052f23829e1180822976108aa4159eb2f4f4
+size 11508560

quantize_config.json

@@ -1,6 +1,6 @@
 {
     "modes": [
-        "q4"
+        "q8"
     ],
     "per_channel": true,
     "reduce_range": true,

tokenizer_config.json

@@ -932,7 +932,7 @@
   "eos_token": "</s>",
   "extra_ids": 100,
   "legacy": true,
-  "model_max_length": 512,
+  "model_max_length": 1000000000000000019884624838656,
   "pad_token": "<pad>",
   "sp_model_kwargs": {},
   "tokenizer_class": "T5Tokenizer",