convert scripts: add feed-forward length for better compatiblilty

This GGUF key is used by all llama.cpp models with upstream support.

gpt4all-backend/scripts/convert_gptj_to_gguf.py

@@ -70,6 +70,7 @@ gguf_writer.add_name("GPT-J")
 gguf_writer.add_context_length(config.n_positions)
 gguf_writer.add_embedding_length(config.n_embd)
 gguf_writer.add_block_count(block_count)
+gguf_writer.add_feed_forward_length(4 * config.n_embd)
 gguf_writer.add_head_count(config.n_head)
 gguf_writer.add_rope_dimension_count(config.rotary_dim)
 gguf_writer.add_layer_norm_eps(config.layer_norm_epsilon)

gpt4all-backend/scripts/convert_mpt_hf_to_gguf.py

@@ -65,6 +65,7 @@ gguf_writer.add_name("MPT")
 gguf_writer.add_context_length(config.max_seq_len)
 gguf_writer.add_embedding_length(config.d_model)
 gguf_writer.add_block_count(block_count)
+gguf_writer.add_feed_forward_length(4 * config.d_model)
 gguf_writer.add_head_count(config.n_heads)
 gguf_writer.add_max_alibi_bias(config.attn_config.alibi_bias_max)
 gguf_writer.add_layer_norm_eps(config.layer_norm_epsilon)

gpt4all-backend/scripts/convert_replit_hf_to_gguf.py

@@ -50,6 +50,7 @@ gguf_writer.add_name("Replit")
 gguf_writer.add_context_length(config.max_seq_len)
 gguf_writer.add_embedding_length(config.d_model)
 gguf_writer.add_block_count(block_count)
+gguf_writer.add_feed_forward_length(4 * config.d_model)
 gguf_writer.add_head_count(config.n_heads)
 gguf_writer.add_max_alibi_bias(config.attn_config.alibi_bias_max)
 gguf_writer.add_layer_norm_eps(config.layer_norm_epsilon)