Fix config for vLLM v2 compatibility: use_cache=true, rope_scaling format
This commit is contained in:
12
config.json
12
config.json
@@ -21,17 +21,15 @@
|
||||
"pad_token_id": 128004,
|
||||
"pretraining_tp": 1,
|
||||
"rms_norm_eps": 1e-05,
|
||||
"rope_parameters": {
|
||||
"tie_word_embeddings": false,
|
||||
"use_cache": true,
|
||||
"vocab_size": 128256,
|
||||
"rope_scaling": {
|
||||
"factor": 8.0,
|
||||
"high_freq_factor": 4.0,
|
||||
"low_freq_factor": 1.0,
|
||||
"original_max_position_embeddings": 8192,
|
||||
"rope_theta": 500000.0,
|
||||
"rope_type": "llama3"
|
||||
},
|
||||
"tie_word_embeddings": false,
|
||||
"unsloth_fixed": true,
|
||||
"unsloth_version": "2026.3.8",
|
||||
"use_cache": false,
|
||||
"vocab_size": 128256
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user