KD-distilled 2-layer student against Llama-2-7B teacher (alpaca-cleaned, 1500 steps, T=2.0, KL loss)

config.json

@@ -5,7 +5,7 @@
   "attention_bias": false,
   "attention_dropout": 0.0,
   "bos_token_id": 1,
-  "dtype": "float16",
+  "dtype": "bfloat16",
   "eos_token_id": 2,
   "head_dim": 128,
   "hidden_act": "silu",
@@ -21,10 +21,12 @@
   "pad_token_id": 0,
   "pretraining_tp": 1,
   "rms_norm_eps": 1e-05,
-  "rope_scaling": null,
-  "rope_theta": 10000.0,
+  "rope_parameters": {
+    "rope_theta": 10000.0,
+    "rope_type": "default"
+  },
   "tie_word_embeddings": false,
-  "transformers_version": "4.56.0",
-  "use_cache": true,
+  "transformers_version": "5.5.0",
+  "use_cache": false,
   "vocab_size": 32000
 }

generation_config.json

@@ -6,5 +6,5 @@
   "pad_token_id": 32000,
   "temperature": 0.9,
   "top_p": 0.6,
-  "transformers_version": "4.56.0"
+  "transformers_version": "5.5.0"
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4428afb8feed1ad5fd25eb8a66eb290a782afd6877509fa8553830d10163b4fc
-size 1333831992
+oid sha256:b27495667b43cc3d89b0dc07f3a85bde18d6aeea2cd1a80dff951de48411a839
+size 1333832016

tokenizer.json

@@ -31,23 +31,13 @@
       "special": true
     }
   ],
-  "normalizer": {
-    "type": "Sequence",
-    "normalizers": [
-      {
-        "type": "Prepend",
-        "prepend": "▁"
-      },
-      {
-        "type": "Replace",
-        "pattern": {
-          "String": " "
-        },
-        "content": "▁"
-      }
-    ]
+  "normalizer": null,
+  "pre_tokenizer": {
+    "type": "Metaspace",
+    "replacement": "▁",
+    "prepend_scheme": "first",
+    "split": false
   },
-  "pre_tokenizer": null,
   "post_processor": {
     "type": "TemplateProcessing",
     "single": [
@@ -129,7 +119,7 @@
   "model": {
     "type": "BPE",
     "dropout": null,
-    "unk_token": "<unk>",
+    "unk_token": null,
     "continuing_subword_prefix": null,
     "end_of_word_suffix": null,
     "fuse_unk": true,

tokenizer_config.json

@@ -1,38 +1,10 @@
 {
-  "add_bos_token": true,
-  "add_eos_token": false,
   "add_prefix_space": null,
-  "added_tokens_decoder": {
-    "0": {
-      "content": "<unk>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": true
-    },
-    "1": {
-      "content": "<s>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": true
-    },
-    "2": {
-      "content": "</s>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": true
-    }
-  },
+  "backend": "tokenizers",
   "bos_token": "<s>",
   "clean_up_tokenization_spaces": false,
   "eos_token": "</s>",
-  "extra_special_tokens": {},
-  "legacy": false,
+  "is_local": false,
   "model_max_length": 1000000000000000019884624838656,
   "pad_token": "<unk>",
   "sp_model_kwargs": {},