upload tiny model

config.json

@@ -0,0 +1,31 @@
+{
+  "_name_or_path": "./",
+  "architectures": [
+    "T5ForConditionalGeneration"
+  ],
+  "d_ff": 16,
+  "d_kv": 6,
+  "d_model": 8,
+  "decoder_start_token_id": 0,
+  "dense_act_fn": "silu",
+  "dropout_rate": 0.0,
+  "eos_token_id": 1,
+  "feed_forward_proj": "gated-silu",
+  "initializer_factor": 1.0,
+  "is_encoder_decoder": true,
+  "is_gated_act": true,
+  "layer_norm_epsilon": 1e-06,
+  "model_type": "t5",
+  "num_decoder_layers": 3,
+  "num_heads": 4,
+  "num_layers": 3,
+  "output_past": true,
+  "pad_token_id": 0,
+  "relative_attention_max_distance": 128,
+  "relative_attention_num_buckets": 64,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.26.0",
+  "use_cache": true,
+  "vocab_size": 256
+}

t5x/README.md

@@ -0,0 +1,3 @@
+---
+license: cc-by-nc-sa-4.0
+---

t5x/checkpoint_0/state.param_states.decoder.decoder.pre_cross_attention_layer_norm.scale.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/state.param_states.decoder.decoder.pre_mlp_layer_norm.scale.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/state.param_states.decoder.decoder.pre_self_attention_layer_norm.scale.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/state.param_states.decoder.decoder.relpos_bias.rel_embedding.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,3,64],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4,3,64],"zarr_format":2}

t5x/checkpoint_0/state.param_states.decoder.decoder_norm.scale.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8],"zarr_format":2}

t5x/checkpoint_0/state.param_states.decoder.logits_dense.kernel.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,256],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,256],"zarr_format":2}

t5x/checkpoint_0/state.param_states.encoder.encoder.pre_attention_layer_norm.scale.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/state.param_states.encoder.encoder.pre_mlp_layer_norm.scale.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/state.param_states.encoder.encoder.relpos_bias.rel_embedding.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,3,64],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4,3,64],"zarr_format":2}

t5x/checkpoint_0/state.param_states.encoder.encoder_norm.scale.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[8],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8],"zarr_format":2}

t5x/checkpoint_0/state.param_states.token_embedder.embedding.v/.zarray

@@ -0,0 +1 @@
+{"chunks":[256,8],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[256,8],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.encoder_decoder_attention.key.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3,4,6],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3,4,6],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.encoder_decoder_attention.out.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,3,6,8],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4,3,6,8],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.encoder_decoder_attention.query.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3,4,6],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3,4,6],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.encoder_decoder_attention.value.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3,4,6],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3,4,6],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.mlp.wi_0.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3,16],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3,16],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.mlp.wi_1.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3,16],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3,16],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.mlp.wo.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[16,3,8],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[16,3,8],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.pre_cross_attention_layer_norm.scale/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.pre_mlp_layer_norm.scale/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.pre_self_attention_layer_norm.scale/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.relpos_bias.rel_embedding/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,3,64],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4,3,64],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.self_attention.key.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[8,3,4,6],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[8,3,4,6],"zarr_format":2}

t5x/checkpoint_0/target.decoder.decoder.self_attention.out.kernel/.zarray

@@ -0,0 +1 @@
+{"chunks":[4,3,6,8],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"<f4","fill_value":null,"filters":null,"order":"C","shape":[4,3,6,8],"zarr_format":2}