Add agent projector for step 6720
Browse files
checkpoint-6720/agent_projector.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:da6dae056d5b7c307324750dcdde113f0b05d8f1852256f4ac76535d9e7e9d73
|
3 |
+
size 267533804
|
checkpoint-6720/config.json
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "lmsys/vicuna-13b-v1.5",
|
3 |
+
"agent_encoder_dim": 1024,
|
4 |
+
"agent_num_encoders": 15,
|
5 |
+
"agent_projector_lr": null,
|
6 |
+
"agent_state_tokens": 64,
|
7 |
+
"architectures": [
|
8 |
+
"LlamaForCausalLM"
|
9 |
+
],
|
10 |
+
"attention_bias": false,
|
11 |
+
"attention_dropout": 0.0,
|
12 |
+
"bos_token_id": 1,
|
13 |
+
"eos_token_id": 2,
|
14 |
+
"hidden_act": "silu",
|
15 |
+
"hidden_size": 5120,
|
16 |
+
"initializer_range": 0.02,
|
17 |
+
"intermediate_size": 13824,
|
18 |
+
"llm_state_tokens": 16,
|
19 |
+
"max_length": 4096,
|
20 |
+
"max_position_embeddings": 4096,
|
21 |
+
"mlp_bias": false,
|
22 |
+
"model_type": "arjun_llama",
|
23 |
+
"num_attention_heads": 40,
|
24 |
+
"num_hidden_layers": 40,
|
25 |
+
"num_key_value_heads": 40,
|
26 |
+
"pad_token_id": 0,
|
27 |
+
"pretraining_tp": 1,
|
28 |
+
"projection_layer": "encoder[\\d]+/ln2",
|
29 |
+
"projector_dropout": 0.1,
|
30 |
+
"projector_hidden_depth": 3,
|
31 |
+
"projector_hidden_dim": 2048,
|
32 |
+
"rms_norm_eps": 1e-05,
|
33 |
+
"rope_scaling": null,
|
34 |
+
"rope_theta": 10000.0,
|
35 |
+
"tie_word_embeddings": false,
|
36 |
+
"tokenizer_model_max_length": 375,
|
37 |
+
"tokenizer_padding_side": "right",
|
38 |
+
"torch_dtype": "float16",
|
39 |
+
"transformers_version": "4.43.1",
|
40 |
+
"tune_agent_adapter": true,
|
41 |
+
"use_cache": false,
|
42 |
+
"use_state_patch_token": true,
|
43 |
+
"use_state_start_end": false,
|
44 |
+
"vocab_size": 32001
|
45 |
+
}
|