feat: Add plumbing for Granite 3.0 8b and 3.1 2b/8b

Branch: GraniteThreeDenseSupport

Signed-off-by: Gabe Goodhart <[email protected]>

Author: gabe-l-hart

torchchat/model_config/models.json

@@ -180,10 +180,31 @@
         "tokenizer_file": "tokenizer.json"
     },
     "ibm-granite/granite-3.0-2b-instruct": {
-        "aliases": ["granite3-2b"],
+        "aliases": ["granite3-2b", "granite3"],
         "distribution_channel": "HuggingFaceSnapshot",
         "distribution_path": "ibm-granite/granite-3.0-2b-instruct",
         "transformer_params_key": "Granite-3.0-2B-Instruct",
         "tokenizer_file": "tokenizer.json"
+    },
+    "ibm-granite/granite-3.0-8b-instruct": {
+        "aliases": ["granite3-8b"],
+        "distribution_channel": "HuggingFaceSnapshot",
+        "distribution_path": "ibm-granite/granite-3.0-8b-instruct",
+        "transformer_params_key": "Granite-3.0-8B-Instruct",
+        "tokenizer_file": "tokenizer.json"
+    },
+    "ibm-granite/granite-3.1-2b-instruct": {
+        "aliases": ["granite3.1-2b", "granite3.1"],
+        "distribution_channel": "HuggingFaceSnapshot",
+        "distribution_path": "ibm-granite/granite-3.1-2b-instruct",
+        "transformer_params_key": "Granite-3.1-2B-Instruct",
+        "tokenizer_file": "tokenizer.json"
+    },
+    "ibm-granite/granite-3.1-8b-instruct": {
+        "aliases": ["granite3.1-8b"],
+        "distribution_channel": "HuggingFaceSnapshot",
+        "distribution_path": "ibm-granite/granite-3.1-8b-instruct",
+        "transformer_params_key": "Granite-3.1-8B-Instruct",
+        "tokenizer_file": "tokenizer.json"
     }
 }

torchchat/model_params/Granite-3.0-8B-Instruct.json

@@ -0,0 +1,20 @@
+{
+    "attention_multiplier": 0.0078125,
+    "embedding_multiplier": 12.0,
+    "dim": 4096,
+    "block_size": 12800,
+    "hidden_dim": 12800,
+    "logits_scaling": 16.0,
+    "n_heads": 32,
+    "n_layers": 40,
+    "n_local_heads": 8,
+    "residual_multiplier": 0.22,
+    "norm_eps": 1e-05,
+    "rope_base": 10000,
+    "tie_word_embeddings": true,
+    "vocab_size": 49155,
+    "use_hf_tokenizer": true,
+    "tokenizer_prepend_bos": false,
+    "attention_bias": false,
+    "feed_forward_bias": false
+}

torchchat/model_params/Granite-3.1-2B-Instruct.json

@@ -0,0 +1,20 @@
+{
+    "attention_multiplier": 0.015625,
+    "embedding_multiplier": 12.0,
+    "dim": 2048,
+    "block_size": 8192,
+    "hidden_dim": 8192,
+    "logits_scaling": 8.0,
+    "n_heads": 32,
+    "n_layers": 40,
+    "n_local_heads": 8,
+    "residual_multiplier": 0.22,
+    "norm_eps": 1e-05,
+    "rope_base": 5000000.0,
+    "tie_word_embeddings": true,
+    "vocab_size": 49155,
+    "use_hf_tokenizer": true,
+    "tokenizer_prepend_bos": false,
+    "attention_bias": false,
+    "feed_forward_bias": false
+}

torchchat/model_params/Granite-3.1-8B-Instruct.json

@@ -0,0 +1,20 @@
+{
+    "attention_multiplier": 0.0078125,
+    "embedding_multiplier": 12.0,
+    "dim": 4096,
+    "block_size": 12800,
+    "hidden_dim": 12800,
+    "logits_scaling": 16.0,
+    "n_heads": 32,
+    "n_layers": 40,
+    "n_local_heads": 8,
+    "residual_multiplier": 0.22,
+    "norm_eps": 1e-05,
+    "rope_base": 10000000.0,
+    "tie_word_embeddings": true,
+    "vocab_size": 49155,
+    "use_hf_tokenizer": true,
+    "tokenizer_prepend_bos": false,
+    "attention_bias": false,
+    "feed_forward_bias": false
+}