added qwen, mistral-v{0.1,0.3}

Ranajoy Sadhukhan · Ranajoy Sadhukhan · commit 628d5d61b4ec · 2024-11-16T15:52:10.000-05:00
diff --git a/Engine/SnapKV/model.py b/Engine/SnapKV/model.py
@@ -75,7 +75,7 @@ def from_name(cls, name: str):
     "Yi-1.5-6b": dict(block_size=4096, n_layer=32, n_head=32, n_local_heads=4, dim=4096, intermediate_size=11008, vocab_size=64000, rope_base=500000.0),
     "Yi-1.5-34b-32k": dict(block_size=32768, n_layer=60, n_head=56, n_local_heads=8, dim=7168, intermediate_size=20480, vocab_size=64000, rope_base=500000.0),
     "Mistral-7B-v0.1": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32000),
-    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768),
+    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768, rope_base=1000000.0),
 }
 
 class KVCache(nn.Module):
diff --git a/Engine/SnapKV/model_draft.py b/Engine/SnapKV/model_draft.py
@@ -75,7 +75,7 @@ def from_name(cls, name: str):
     "Yi-1.5-6b": dict(block_size=4096, n_layer=32, n_head=32, n_local_heads=4, dim=4096, intermediate_size=11008, vocab_size=64000, rope_base=500000.0),
     "Yi-1.5-34b-32k": dict(block_size=32768, n_layer=60, n_head=56, n_local_heads=8, dim=7168, intermediate_size=20480, vocab_size=64000, rope_base=500000.0),
     "Mistral-7B-v0.1": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32000),
-    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768),
+    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768, rope_base=1000000.0),
 }
 
 class KVCache(nn.Module):
diff --git a/Engine/StreamingLLM/model.py b/Engine/StreamingLLM/model.py
@@ -75,7 +75,7 @@ def from_name(cls, name: str):
     "Yi-1.5-6b": dict(block_size=4096, n_layer=32, n_head=32, n_local_heads=4, dim=4096, intermediate_size=11008, vocab_size=64000, rope_base=500000.0),
     "Yi-1.5-34b-32k": dict(block_size=32768, n_layer=60, n_head=56, n_local_heads=8, dim=7168, intermediate_size=20480, vocab_size=64000, rope_base=500000.0),
     "Mistral-7B-v0.1": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32000),
-    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768),
+    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768, rope_base=1000000.0),
 }
 
 class KVCache(nn.Module):
diff --git a/Engine/StreamingLLM/model_draft.py b/Engine/StreamingLLM/model_draft.py
@@ -75,7 +75,7 @@ def from_name(cls, name: str):
     "Yi-1.5-6b": dict(block_size=4096, n_layer=32, n_head=32, n_local_heads=4, dim=4096, intermediate_size=11008, vocab_size=64000, rope_base=500000.0),
     "Yi-1.5-34b-32k": dict(block_size=32768, n_layer=60, n_head=56, n_local_heads=8, dim=7168, intermediate_size=20480, vocab_size=64000, rope_base=500000.0),
     "Mistral-7B-v0.1": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32000),
-    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768),
+    "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768, rope_base=1000000.0),
 }
 
 class KVCache(nn.Module):

Original file line number	Diff line number	Diff line change
`@@ -75,7 +75,7 @@ def from_name(cls, name: str):`
`75`	`75`	`"Yi-1.5-6b": dict(block_size=4096, n_layer=32, n_head=32, n_local_heads=4, dim=4096, intermediate_size=11008, vocab_size=64000, rope_base=500000.0),`
`76`	`76`	`"Yi-1.5-34b-32k": dict(block_size=32768, n_layer=60, n_head=56, n_local_heads=8, dim=7168, intermediate_size=20480, vocab_size=64000, rope_base=500000.0),`
`77`	`77`	`"Mistral-7B-v0.1": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32000),`
`78`		`- "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768),`
	`78`	`+ "Mistral-7B-v0.3": dict(n_layer=32, n_head=32, n_local_heads=8, dim=4096, intermediate_size=14336, vocab_size=32768, rope_base=1000000.0),`
`79`	`79`	`}`
`80`	`80`
`81`	`81`	`class KVCache(nn.Module):`